Appendix C Mass Balance Modeling Analysis - Passaic River ...

Appendix C

Mass Balance Modeling Analysis

Appendix C: Mass Balance Modeling Analysis Lower Eight Miles of the Lower Passaic River 2014

LOWER EIGHT MILES OF THE LOWER PASSAIC RIVER APPENDIX C: MASS BALANCE MODELING ANALYSIS

TABLE OF CONTENTS

1 Introduction .......................................................................................................... 1-1

2 Overview of the Fate and Transport analysis ...................................................... 2-1

2.1 Contaminant Mass Balance Considerations for the Lower Passaic River ... 2-3

2.2 Forecasting Contaminant Concentrations in Surface Sediments ................. 2-4

3 Empirical Mass Balance for the Lower Passaic River ......................................... 3-1

3.1 General Summary of Model ......................................................................... 3-1

3.2 Model Formation ......................................................................................... 3-5

3.3 Identifying Contaminants for Inclusion in EMB Model .............................. 3-9

3.4 Best Estimate Scenario .............................................................................. 3-11

3.5 Sensitivity Analysis of SWO Concentrations and Solids Constraint ........ 3-12

3.6 Monte Carlo Analysis of Uncertainty and Variability in Contaminant

Concentrations ....................................................................................................... 3-13

3.7 Model Performance Evaluation ................................................................. 3-14

3.8 Model Limitation ....................................................................................... 3-15

4 Modeling Results ................................................................................................. 4-1

4.1 EMB Model Solids Balance Results: Best Estimate Scenario and Uncertainty

4-1

4.2 EMB Model Contaminant Fate and Transport ............................................ 4-2

4.3 Evaluation of EMB Model Performance ..................................................... 4-9

4.4 Sensitivity Analysis ................................................................................... 4-10

4.5 Assessment of the Cooperating Parties Group (CPG) 2008-2009 Data .... 4-11

5 Forecasting Contaminant Concentrations ............................................................ 5-1

5.1 Development of Concentration Half Times for the Excess Passaic River

Sediment Burden ...................................................................................................... 5-3

5.2 Forecasts of Sediment Concentrations for FFS Remedial Alternatives ..... 5-10

5.3 Incorporation of the CPG 2008-2009 Data ................................................ 5-22

6 Summary .............................................................................................................. 6-1


i

6.1 Summary of EMB Results ........................................................................... 6-1

6.2 Summary of Contaminant Forecast Results ................................................. 6-2

7 Acronyms ............................................................................................................. 7-1

8 References ............................................................................................................ 8-1


ii


LIST OF TABLES

Table 3-1 EMB Model Optimization Parameters

Table 3-2 Additional Parameters Used for EMB Model Evaluation

Table 3-3 Average Resuspension (Lower Passaic River) Concentrations for

Selected Contaminants (0-6 inch Surface Sediment Samples)

Table 3-4 Newark Bay Northern End and Southern End Average Concentrations

for Selected Contaminants

Table 3-5 Upper Passaic River Recently-Deposited Surface Sediment

Concentrations for Selected Contaminants

Table 3-6 Tributary Average Concentrations for Selected Contaminants

Table 3-7 Average CSO and SWO Concentrations for Selected Contaminants

Table 3-8 Average Lower Passaic River Recently-Deposited (Be-7 Bearing)

Sediment Concentrations for Selected Contaminants

Table 4-1 Contaminant Burden Attributed to Resuspension

Table 4-2 Summary of Solids Contribution Results for Best Estimate Scenarios

Table 4-3 Summary of Solids Contribution Results for SWO Sensitivity Scenarios

Table 4-4 Summary of Solids Contribution Results for Relaxed Solids Constraint

Sensitivity Scenarios

Table 4-5 Comparison of 2008 Tributary and Dundee Dam Measured Surface

Sediment Concentrations with Monte Carlo Simulated Model Inputs

Table 5-1 Comparison of the Average 1980s Concentrations and 2005 Surface

Sediment Concentrations for Selected Contaminants

Table 5-2 Contaminant Half Times for “Excess Concentration” in Lower Passaic

River Sediments Based on High Resolution Cores from 2005 and

Surface Samples in 2007

Table 5-3a Reduction in Best Estimate Forecasted Surface Sediment Concentrations

in 2059 relative to 2017


iii

Table 5-3b Best Estimate Forecasted Surface Sediment Concentrations in 2017 and

2059


iv


LIST OF FIGURES

Figure 3-1 Sources of Model for the Lower Passaic River Empirical Mass Balance

Figure 3-2 Relationship between Average Contaminant Concentration and Standard

Error for the Recently-Deposited Sediments in the Main Stem Lower

Passaic River

Figure 3-3 Cluster Analysis for PAHs

Figure 3-4 Cluster Analysis for Metals

Figure 4-1 Best Estimate of Relative Solids Contribution to the Lower Passaic

River Estimated by the EMB Model

Figure 4-2 Fractional Contribution for Solids (Results from Monte Carlo Analysis

and Best Estimate Solution)

Figure 4-3 Source Concentration and Mass Balance for 2,3,7,8-TCDD

Figure 4-4 Source Concentration and Mass Balance for Total Tetra-dioxins

Figure 4-5 Source Concentration and Mass Balance for Total PCBs

Figure 4-6 Source Concentration and Mass Balance for Benzo[a]pyrene

Figure 4-7 Source Concentration and Mass Balance for Fluoranthene

Figure 4-8 Source Concentration and Mass Balance for 4,4’-DDE

Figure 4-9 Source Concentration and Mass Balance for gamma-Chlordane

Figure 4-10 Source Concentration and Mass Balance for Copper

Figure 4-11 Source Concentration and Mass Balance for Chromium

Figure 4-12 Source Concentration and Mass Balance for Mercury

Figure 4-13 Source Concentration and Mass Balance for Lead

Figure 4-14 Source Concentration and Mass Balance for Iron

Figure 4-15 Source Concentration and Mass Balance for Total Organic Carbon

Figure 4-16 Dieldrin Contribution to the Lower Passaic River

Figure 4-17 Phenanthrene Contribution to the Lower Passaic River

Figure 4-18 Evaluation of EMB Model Performance


v

Figure 5-1 Excess 2,3,7,8-TCDD Concentration vs. Approximate Year of

Deposition

Figure 5-2 Excess Mercury Concentration vs. Approximate Year of Deposition

Figure 5-3 Excess Lead Concentration vs. Approximate Year of Deposition

Figure 5-4 Excess Copper Concentration vs. Approximate Year of Deposition

Figure 5-5 Excess 4,4’-DDE Concentration vs. Approximate Year of Deposition

Figure 5-6 Excess gamma-Chlordane Concentration vs. Approximate Year of

Deposition

Figure 5-7 Excess Dieldrin Concentration vs. Approximate Year of Deposition

Figure 5-8 Excess Total PCBs Concentration vs. Approximate Year of Deposition

Figure 5-9 Excess High Molecular Weight PAH Concentration vs. Approximate

Year of Deposition

Figure 5-10 Excess Low Molecular Weight PAH Concentration vs. Approximate

Year of Deposition

Figure 5-11 2,3,7,8-TCDD Best Estimates of Trajectories for 0-6 inch Biologically

Active Layer

Figure 5-12 Mercury Best Estimates of Trajectories for 0-6 inch Biologically Active

Layer

Figure 5-13 Lead Best Estimates of Trajectories for 0-6 inch Biologically Active

Layer

Figure 5-14 Copper Best Estimates of Trajectories for 0-6 inch Biologically Active

Layer

Figure 5-15 4,4’-DDE Best Estimates of Trajectories for 0-6 inch Biologically

Active Layer

Figure 5-16 gamma-Chlordane Best Estimates of Trajectories for 0-6 inch

Biologically Active Layer

Figure 5-17 Total PCB Best Estimates of Trajectories for 0-6 inch Biologically

Active Layer

Figure 5-18 HMW PAH Best Estimates of Trajectories for 0-6 inch Biologically Active Layer


vi

LOWER EIGHT MILES OF THE LOWER PASSAIC RIVER

APPENDIX C: MASS BALANCE MODELING ANALYSIS

LIST OF ATTACHMENTS

Attachment A Monte Carlo Methodology for Uncertainty Analysis on the EMB

Model and Forecast Trajectories

Attachment B Estimating the Common Half Time for Legacy Sediments in

Lower Passaic River

Attachment C Derivation of the Trajectories


vii

1 INTRODUCTION

This appendix describes the Empirical Mass Balance (EMB) modeling analysis

developed to support the Focused Feasibility Study (FFS) of the lower eight miles of the

Lower Passaic River1. It encompasses sources to and receptors present in the tidal portion

of the river, from River Mile (RM2) 0 to RM17.4, to provide a more complete

understanding of contaminant fate and transport. The appendix is composed of the

following chapters in addition to the introduction:

• Chapter 2, Overview of the Fate and Transport Conceptual Analysis: provides an

overview of the contaminant fate and transport conceptual models.

• Chapter 3, Empirical Mass Balance Model for the Lower Passaic River: describes the

EMB model established for the river, which is designed to characterize the fate and

transport of contaminants in the Lower Passaic River.

• Chapter 4, Empirical Mass Balance Model Results: presents the results of the EMB

for contaminants and solids.

• Chapter 5, Forecasting Contaminant Concentrations: presents the forecast

concentrations of contaminants in Lower Passaic River surface sediment based on the

EMB results.

• Chapter 6, Summary: summaries the results of the EMB and future forecast of

contaminant concentrations.

• Chapter 7, Acronyms: defines the acronyms used in this appendix.

• Chapter 8, References: lists the references used in this appendix.

Appendix A (Data Evaluation Reports) and Appendix C (Mass Balance Modeling

Analysis) contain elements previously discussed in the Draft Comprehensive Conceptual

1 Throughout this appendix, the term “Lower Passaic River” is used to refer to the tidal portion of the Passaic River, from Dundee Dam to the river mouth at Newark Bay (RM0 to RM17.4). The term “lower 8 miles” refers to the FFS Study Area, from RM0 to RM8.3. The term “Upper Passaic River refers to the freshwater portion of the Passaic River above Dundee Dam. 2 The FFS uses the “River Mile” (RM) system developed by the United States Army Corps of Engineers (USACE), which follows the navigation channel of the Lower Passaic River. The Data Evaluation Reports (Appendix A), Empirical Mass Balance (Appendix C) and Lower Passaic River-Newark Bay model (Appendix B) were initially developed at the beginning of the 17-mile Remedial Investigation and Feasibility Study (RI/FS), and thus follow a RM system developed for that RI/FS, which follows the geographic centerline of the river. RM0 is defined by an imaginary line between two marker lighthouses at the confluence of the Lower Passaic River and Newark Bay: one in Essex County just offshore of Newark and the other in Hudson County just offshore of Kearny Point. River miles then continue upriver to the Dundee Dam (RM17.4). The two RM systems are about 0.2 miles apart.

Appendix C: Mass Balance Modeling Analysis Lower Eight Miles of the Lower Passaic River 2014 1-1

Site Model (Malcolm Pirnie, 2008). A contractor-led independent peer review of this

document was conducted from May 31 through July 7, 2008. As a result of this peer

review several changes were made and were incorporated into Appendices A and C:

• A Monte Carlo technique was used to estimate uncertainty in the empirical mass

balance and the prediction of future sediment concentrations under various

remedial scenarios. The results were incorporated into Appendix C.

• Additional sampling events were conducted in 2008 to collect a set of low

resolution cores above RM8 and a set of suspended solids samples from the CSOs

and SWOs to address acknowledged data gaps. Also, in addition to the use of

Monte Carlo analysis to estimate uncertainties, the discussion of the high

resolution core dating assignments was expanded and refined. The results were

incorporated into Appendices A and C.


2 OVERVIEW OF THE FATE AND TRANSPORT ANALYSIS

For the FFS, two separate model-based examinations of contaminant transport were

conducted. This Appendix presents one of these examinations, called the EMB Model,

which used an empirical receptor modeling approach to simultaneously examine the

particle-borne concentrations of a broad suite of contaminants and other compounds to

establish the magnitude of each contaminant contribution from each of the major sources

to the estuary. Appendix B3 presents the other examination, the Lower Passaic River-

Newark Bay Model, which used a mechanistic modeling approach, incorporating

hydrodynamic and sediment transport modeling results while modeling various

contaminants on an individual basis.

In this Appendix, the goal of the modeling was to infer contaminant contributions from

various sources, and to use this result to empirically forecast future concentrations under

different remedial alternatives. To do this for the Lower Passaic River, a “receptor”

modeling approach was undertaken. Receptor models are empirically-based, focus on the

behavior at the receptor site, and infer contributions from different sources based on

multivariate measurements taken at the receptor site and likely sources.

Receptor models have been widely used in the field of air pollution [e.g., United States

Environmental Protection Agency (USEPA) Chemical Mass Balance (CMB) Model

(Watson et al., 2004)] as tools for identification of pollutant sources and evaluation of

their relative contributions. Recently, receptor models have also been applied to sediment

sites that are contaminated with polychlorinated biphenyl (PCB), polychlorinated

dibenzodioxin/furan (PCDD/F), and polycyclic aromatic hydrocarbon (PAH) compounds.

Examples of these sediment contamination sites include: the Fox River in Wisconsin (Su

et al., 2000), San Francisco Bay in California (Johnson et al., 2000), the Ashtabula River

in Ohio (Imamoglu et al., 2002), Lake Calumet in Chicago (Bzdusek et al., 2004), and

Tokyo Bay and Lake Shinji in Japan (Ogura et al., 2005). The objectives of the receptor

3 This appendix makes extensive use of cross references to direct the reader to the sources of the analyses and conclusions incorporated in this appendix.


model are to determine the number of sources contributing to the system, the contaminant

composition of each source, and the relative contribution of each source to the receptor

site.

For the FFS, the receptor model described here will focus on explaining the contaminant

concentrations in recently-deposited sediments [i.e., Beryllium-7 (Be-7)4 -bearing

sediment] in the Lower Passaic River. Recently-deposited sediments integrate the various

sources to the Lower Passaic River water column, as well as internal river processes that

affected these sediments when they were deposited during the prior six to twelve month

period. Because the source compositions are known and data are available to determine

their contaminant composition, the non-negative constrained contaminant mass balance

approach is used. This approach used in the analysis follows a recent application of the

USEPA CMB model that was combined with Monte Carlo techniques5, to account for

uncertainty and variability in the data (Ogura et al., 2005). A detailed description of the

Monte Carlo analysis methodology and how it was used to account for uncertainties in

source and receptor compositions is given in Attachment A.

The following sections describe the empirical modeling analyses that were incorporated

in the development of the Conceptual Site Model (CSM) to gain insight into some of the

important environmental processes occurring in the Lower Passaic River. The analyses

performed included: Contaminant Mass Balance for the Lower Passaic River and

Development of a Mass Balance Forecast Model to forecast contaminant concentrations

for the Lower Passaic River.

4 Be-7 is a naturally occurring, particle-reactive radioisotope with a short half-life (53 days). The presence of Be-7 in surface sediments suggests that the associated solids were deposited on the sediment bed within the last 6 months (termed “recently-deposited surface sediments”) prior to collection. 5 Monte Carlo is an analytical technique where a large number of simulations are run, using randomly selected quantities from a specified distribution for each variable, and the output then reviewed and evaluated to determine which values are the most likely. In this Monte Carlo simulation, the concentrations of contaminants in the sources and receptor are generated randomly from defined distributions, and the mass balance calculation is repeated many times with different randomly determined data to allow statistical conclusions to be drawn.


2.1 Contaminant Mass Balance Considerations for the Lower Passaic River

Contaminants are transmitted through the environment by a variety of processes,

including advection and dispersion, as both dissolved constituents and adsorbed

constituents of particles. The contaminants themselves also undergo alterations due to

environmental processes such as adsorption to and desorption from particles and

degradation through microbial respiration. Contaminant fate and transport analysis

attempts to understand the effects of these processes either through mechanistic or

empirical means. For the FFS, both means were used to provide two lines of evidence on

which to base decisions. The mechanistic contaminant fate and transport model is

presented in Appendix B. The EMB model, a semi-empirical formulation presented here,

evaluates the relative contributions of the important boundary conditions [the Upper

Passaic River, Newark Bay, tributaries, Combined Sewer Overflows/Stormwater Outfalls

(CSOs/SWOs) and resuspended legacy sediment acting as sources to the recently

depositing sediments (i.e., Be-7-bearing sediment)]. Note that the term “resuspension of

legacy sediment” represents all the net sediment transfer processes from the bed of the

Lower Passaic River that will affect recently-deposited sediment, including:

resuspension, porewater exchange, and bioturbation. The EMB model for the Lower

Passaic River is developed in Chapter 3, with results and conclusions provided in Chapter

4.

The following tasks were conducted to prepare and solve the EMB model:

• A contaminant mass balance equation was developed to determine the relative

contribution of each external source of fine-grained solids and associated

contaminants (Upper Passaic River, tributaries, CSOs/SWOs, and Newark Bay) to the

recently-deposited (Be-7-bearing) sediments of the Lower Passaic River.

• An empirically-based receptor model was selected to solve the mass balance

equations for the relative contributions of the known sources to the receptor. The

model combines a non-negative constrained contaminant mass balance with

sensitivity analysis simulations to address variability and uncertainty in the source

characterizations.


• The solids contribution from the tributaries and point discharges were further

constrained using their watershed areas to ensure that their model-estimated solids

contributions do not exceed their watershed solids carrying capacity.

• Contaminant parameters from the available datasets were subjected to a cluster

analysis to identify independent contaminants that were uniquely associated with the

sources. The Lower Passaic River accumulates solids that originate from several

sources. In order for the EMB model to decipher the contribution of these sources to

the receptor sediments, independent parameters must be identified and applied in the

model. Independent parameters are contaminants that have independent sources, or

different fate and transport processes, or both. The combination of contaminants

selected for analysis must provide a unique pattern for each of the various sources in

order for a unique solution to be obtained by the model.

• A total of 22 parameters were used in the model. Of these, 13 were directly used in

model optimization to determine the solids contributions. The remaining nine were

used to further evaluate the model performance.

• Model performance was evaluated using a normalized mean error defined as the

difference between the predicted and the observed, normalized to the observed

receptor concentration for each parameter.

• Uncertainties in source and receptor composition and spatial variability in

contaminant concentrations were accounted for through a Monte Carlo analysis.

2.2 Forecasting Contaminant Concentrations in Surface Sediments

Using the results of the EMB model, a two-layer single box model was developed for use

in forecasting Lower Passaic River contaminant concentrations in sediment. This is

described in Chapter 5. The average surface concentrations for various contaminants in

the 0 to 6 inch sediment layer of the Lower Passaic River were empirically forecast under

the four remedial alternatives being evaluated in the FFS using a numerical model

combined with a stochastic simulation. The forecasting formulation aggregates the river

section between RM2 to RM12 as a two-layer single box model consisting of a water

column where mixing of particles from external sources and resuspension occurs, and a


mixed-layer surface sediment bed to which particle deposition from the water column

occurs. The rationale for using the single aggregate representation of this river section

follows from observations of recently-deposited sediments which show little longitudinal

variation in concentrations from RM2 to RM12 (see Data Evaluation Report No. 4 in

Appendix A ). Note that there are concentration gradients at either end of this river

section representing mixing zones with Upper Passaic River solids (i.e., from RM12 to

RM17.4) and Newark Bay solids (from RM0 to RM2), each with relatively low

contaminant concentrations. Furthermore, although the 1995 Tierra Solutions (TSI)

surface sediment data (see Data Evaluation Report No. 1 in Appendix A) indicate

significant spatial variability in surface contaminant concentrations in the river, this

variability (as well as other sources of variability) were accounted for stochastically by a

Monte Carlo simulation approach, providing an estimate of the distribution of future

contaminant concentrations in the river bed. The forecasting analysis integrated results

from the EMB model (Section 3.0), observed surface sediment concentrations (Data

Evaluation Report No. 4 in Appendix A), current contaminant compositions of external

sources (Data Evaluation Report No. 2 in Appendix A), and historical trends of

contamination from dated sediment cores (Data Evaluation Report No. 3 in Appendix A).


3 EMPIRICAL MASS BALANCE FOR THE LOWER PASSAIC

RIVER

3.1 General Summary of Model

Understanding the various contaminant inputs to the river is essential to determining the

effectiveness of remedial strategies. For this reason, it is necessary to establish the

importance of each potential source of contaminants to the Lower Passaic River. The

EMB model was developed to estimate the magnitude of the tributaries, CSOs/SWOs,

Newark Bay, and Upper Passaic River as contaminant sources relative to the

resuspension of legacy sediments and their associated contaminant inventory (Figure 3-

1), in order to aid decision-making regarding the remedial alternatives being evaluated in

the FFS.

As part of the process to evaluate alternatives, the FFS requires an estimation of the post-

remediation contaminant concentrations for each alternative. The FFS also requires an

estimation of the potential risk from exposure to these future contaminant concentrations.

Before post-remediation surface sediment concentrations can be predicted, the current

conditions in the river must be understood and the relative contaminant burden currently

delivered from each source to the Lower Passaic River must be quantified. As shown on

Figure 3-1, the recently-deposited sediment concentrations in the Lower Passaic River are

derived from some combination of several sources, which can be represented with the

following contaminant mass balance equation for each contaminant (i) (Equation 3-1):

total

iRSP

iCSO

iSWOR

iR

iSR

iNB

iDDi

surface SMMMMMMMC ++++++

= /23 Equation 3-1

Where

Cisurface: contaminant i concentration in the Lower Passaic River surface sediments


MiDD: contaminant i mass derived from the Upper Passaic River (The subscript

DD is a reference to the Dundee Dam, the structure that divides the Lower

and Upper Passaic Rivers.)

MiNB: contaminant i mass derived from Newark Bay

MiSR: contaminant i mass derived from Saddle River

Mi3R: contaminant i mass derived from Third River

Mi2R/SWO: contaminant i mass derived from Second River and the SWOs

MiCSO: contaminant i mass derived from the CSOs

MiRSP: contaminant i mass derived from sediment resuspension

STotal: total sediment mass load deposited in the Lower Passaic River

Note that the phrasing “derived from” indicates that the mass contribution comes from a

specific source, but not all of the mass delivered by these sources is deposited on the

surface of the sediment bed of the Lower Passaic River. Equation 3-1 represents the

recently-deposited surface sediments of the Lower Passaic River as a combination of the

solids and contaminant mass originating from various sources. Based on this contaminant

mass balance, a receptor6-type model was developed where the total contaminant mass

present in the sediments of the receptor (i.e., the recently-deposited, Be-7-bearing

sediments in the Lower Passaic River) is the sum of the mass contributions from the

individual sources. For a fixed number of sources (p), the receptor observation of the ith

contaminant (i = 1, 2 …, j) is modeled as a linear combination of sources’ contaminant

species as presented in Equation 3-2. (Equation 3-2 is an algebraic manipulation of

Equation 3-1 where the contaminant mass from each source is represented by a

concentration and a solids fraction.)

∑=

+=p

jiijji XfY

1ε Equation 3-2

6 The term “receptor” is used throughout Chapter 3 of this appendix to refer to the concentrations in sediments depositing on the river bottom (i.e., recently -deposited sediments). This receptor represents the integration of the various external and internal loads. This term is not the same as the risk assessment definition of the term, as used elsewhere in the FFS.


Where

Yi: receptor concentration for the ith contaminant concentration

Xij: the ith contaminant concentration for the jth source

fi: fraction of solids contributed by the jth source to the receptor

ei: error associated with the concentration of the ith contaminant

p: number of sources

Note that the term fi represents the fraction of solids by the ith source to the Be-7-bearing

sediments (i.e., the receptor). Given that there are seven possible sources, there are then

seven fi terms. The regression process solves for these seven fi terms by optimizing the fi

values and minimizing the residual error term ei. The EMB model is designed to be

solved simultaneously for the contaminant burden of the ith contaminant species for each

jth source, assuming that the model parameters are independent. The following premises

were considered in the design of the EMB model:

• The number of sources is known and includes the Upper Passaic River (above

Dundee Dam), Saddle River, Third River, Second River, CSOs, SWOs, resuspension

of legacy sediments within the Lower Passaic River, and Newark Bay. Contaminant

inputs from atmospheric deposition and groundwater have been determined to be

negligible [See Data Evaluation Report No. 2 in Appendix A].

• Because the SWO samples were collected from points below the high-tide mark,

solids collected from the SWOs represent a mixture of river-originated solids and

SWO-originated solids. Since the data from the SWO samples were compromised by

the intrusion of Lower Passaic River sediments into the SWOs, the contribution from

Second River and the SWOs was combined as a single term in the model and the

contaminant characteristics of both were based on samples taken in Second River (see

Data Evaluation Report No. 2 in Appendix A for a discussion of SWO data quality).

Second River was deemed to be representative of SWO discharges into the Passaic

River, because the Second River drains a highly-urbanized watershed that is fed

primarily by storm water collection systems. A sensitivity analysis simulation was

conducted to evaluate the impact of the SWO concentrations on model results.


• The nature of the sources is known, and the available data represents the current

average composition of all these sources. In most instances, those sources are

characterized by samples collected at or near their discharge points to the Lower

Passaic River. The source characteristics for resuspension of Lower Passaic River

sediments were represented by the surface concentrations from the 1995 TSI

dataset. The 1995 TSI dataset is considered representative of the contaminant

signature of the net transfer of sediment from the bed to the water column through

mechanisms such as erosion, bioturbation, and other resuspension processes.

Although the surface sediment concentration in the 1995 TSI data sets were used

to define the resuspension signature for the EMB model, this analysis does not

assume that erosion is limited to the surface sediments only. The concentrations

of most of the contaminants analyzed in the EMB model vary by several orders of

magnitude in the 1995 TSI surface sediment data (see Data Evaluation Report No

4 in Appendix A). For example, surface concentrations of 2,3,7,8-

tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD) vary by four orders of magnitude

(see Data Evaluation Report No. 4 in Appendix A). This variability likely

represents the range of concentrations of sediments that are resuspended into the

water column, which is incorporated into the EMB model through a Monte Carlo

analysis. Note that median surface sediment contaminant concentrations have not

changed much between 1995 to 2012 (see Temporal and Spatial Trends sections

of Data Evaluation Report No. 4 in Appendix A). Furthermore, the 1995

Remedial Investigation (RI) program was designed to follow a systematic (i.e.,

unbiased) sampling scheme. Sediment cores were collected from multiple

transects spaced at quarter mile intervals, with three cores along each transect (see

Figure 2.1-1 of Data Evaluation Report No. 4 in Appendix A).

• The model focuses on the movement of solids; therefore, it tracks the contaminant

species associated with the solids. Since the modeled compounds are primarily

hydrophobic contaminants, dissolved-phase concentrations (and the processes

impacting dissolved-phase concentrations) are relatively small and are not addressed

by the model.


• The contaminant species included in the mass balance do not react with each other

and can be added linearly.

• The EMB model system is over-determined [there are 13 parameters (twelve

contaminants plus Total Organic Carbon (TOC), see Table 3-1) and 7 equations],

meaning that the number of sources is less than or equal to the number of

contaminant species. Because it is over-determined, several physical constraints were

applied to guide the model solution (see sections 3.2.1, 3.5, and 4.4).

• The source profiles [i.e., the relative proportion of the 13 parameters (see Table 3-1)

in each source] are linearly independent of each other, and any contaminant

transformations or losses that occur between the source and receptor are not

considered. Only contaminants that aid in differentiating among the sources (i.e.,

make the sources independent) were selected for the modeling analysis.

• Uncertainties in the measurement of contaminants and spatial variability are

addressed through a Monte Carlo simulation approach.

Once the receptor solids and source solids were characterized, statistically independent

parameters were identified (see Section 3.3) and the average concentrations of these

parameters were used as inputs to the EMB model. The output of the EMB model

quantifies the relative contribution of the contaminant burden and solids load from each

source to the recently-deposited (Be-7-bearing) Lower Passaic River sediment. The fate

and transport implications of the model output were then described qualitatively for each

contaminant. This modeling approach, which was used to describe the contaminant

burden of the river under current conditions, was also used to provide insight to the

application of the mechanistic model described in Appendix B.

3.2 Model Formation

3.2.1 Function and Constraints / Assumptions and Limitations

The receptor model was formulated following the principles described in Section 3.1 and

using Equations 3-1 and Equation 3-2. The linear equations generated from Equation 3-2

were solved simultaneously using a least square solution to determine the fraction of the


contaminant burden (i.e., the contaminant flux) contributed by each source to the Lower

Passaic River. This solution was achieved by establishing an objective function as

defined by Soonthornnonda and Christensen (2008) below (Equation 3-3):

( ) ( ){ }∑

∑

∑

=

=

=

+

−

=n

ip

jijijik

p

jijji

XerfYer

XfYQ

1

1

22

2

12

.... Equation 3-3

Where:

Q2: weighted sum of squares differences between predicted and observed

receptor concentrations

Yi: concentration in Lower Passaic River surface sediment for the ith

contaminant

fj: fraction of solids contributed by the jth source to the Lower Passaic River

Xij: ith contaminant concentration from the jth source

p: number of sources

n: number of contaminant species (assuming that n > p)

r.e.k: relative error or uncertainty and spatial variability in Yi.

r.e.i: relative error or uncertainty and spatial variability in Xij.

To optimize the fj values, the objective is to choose the fj values so as to minimize the

value of Q2. According to Soonthornnonda and Christensen (2008), these relative errors

can be characterized by the standard error of the measurements for each contaminant and

Equation 3-3 reduces to an expression used by Ogura et al., (2005) given by:

2

1 1

2 1∑ ∑= =

−=

n

i

p

jijji

i

XfYQσ

Equation 3-4

Where:


σi: uncertainty and spatial variability determined by the standard error in

contaminant concentrations.

Consistent with Ogura et al., (2005), the uncertainty or standard error term σi in this

analysis is replaced by Yi itself in the objective function because the magnitude of the

variability was found to depend on the magnitude of the detected concentration (Figure 3-

2). Dioxins/Furans, which have the smallest concentrations, have the smallest standard

errors, while the heavy metals, which have the highest concentrations, have the highest

standard errors. Without consideration of these differences, the chemicals with the largest

variability will dominate the calculation.

The solution of the objective function (Equation 3-4) was limited by the following

constraints:

• The sum of the solids fractions contributed by each source (fj) equals one. (This

constraint was tested in a sensitivity analysis on the model solution.)

• Non-negativity constraint is applied to ensure that a source cannot have a negative

contribution: fj > 0 (i.e., no source can subtract contamination from the Lower Passaic

River sediments).

• A watershed delivery constraint is applied to avoid solids contribution results from

the least squares equation that are unrealistic with regard to the delivery capacity of

the sources. These constraints were written for the inputs from tributaries (Saddle

River, Second River/SWOs, and Third River) and CSOs as limiting linear functions

of contribution from the Upper Passaic River using a tolerance of ±50 percent of the

watershed area ratios according to Equation 3-4 (see Table 5-1 in Data Evaluation

Report No. 2 in Appendix A for watershed areas). Note that the mass balance is not

contingent on the absolute magnitude of the solids load or watershed area but only on

the relative proportions of each source. The watershed delivery constraints are

expressed as a fraction of the solids load delivered by the Upper Passaic River as

follows:


111.0037.0 _ ≤≤RiverPassaicUpper

RiverSaddle

SS

Equation 3-4a

024.0008.0 _ ≤≤RiverPassaicUpper

RiverThird

SS

Equation 3-4b

061.0020.0 /_ ≤≤RiverPassaicUpper

SWORiverSecond

SS

Equation 3-4c

046.0015.0 ≤≤RiverPassaicUpper

CSO

SS

Equation 3-4d

Where

SSaddle_River: solids load from the Saddle River

SThird_River: solids load from the Third River

SSecond_River/SWO: solids load from the Second River and SWOs

SCSO: solids load from CSOs

SUpper Passaic River: solids load from the Upper Passaic River

The EMB model calculations were performed using a combination of Microsoft Excel®

Solver and the Crystal Ball® 7 (Decisioneering, Denver, CO, USA) add-on for Microsoft

Excel® (a tool typically used for solving optimization problems). Using the model

formulation described above, a best estimate solution was obtained based on the average

source and receptor concentrations. A Monte Carlo analysis consisting of 10,000

iterations of randomly generated source and receptor contaminant concentrations was

performed to assess the impact of variability and uncertainty in source and receptor

concentrations on the best estimate solution. Finally, sensitivity analysis simulations were

conducted to evaluate the impact of the SWO concentrations and the model solids

constraint on the best estimate model solution. The model best estimate solution was

assessed using model performance criteria (described below).


3.3 Identifying Contaminants for Inclusion in EMB Model

The Lower Passaic River accumulates solids that originate from several sources. In order

for the EMB model to decipher the contribution of these sources to the receptor

sediments, independent parameters must be identified and applied in the model.

Independent parameters are contaminants that have independent sources and/or different

fate and transport processes. Note that in the special case where a contaminant is not

independent of another contaminant, but together they form a fingerprint that can be used

to distinguish the sources, the two contaminants can be considered in the analysis. The

combination of contaminants selected for analysis must provide a relatively unique

pattern for each of the various sources in order for a unique solution to be obtained by the

model. Contaminants were selected from each of the compound classes, including:

dioxins/furans, PCBs, PAHs, pesticides, and metals. The individual contaminants chosen

are as follows:

• For dioxin/furan compounds, 2,3,7,8-TCDD and total tetrachlorodibenzodioxin

(Total TCDD) were selected. Although they are not independent parameters, both

were included because their ratio is an important tracer for Lower Passaic River

solids throughout the New York-New Jersey Harbor Estuary (Chaky, 2003).

• For PCBs, the data for the external sources were reported on a congener basis.

However, because the TSI 1995 data was reported on an Aroclor basis, the sum of

PCB Aroclors was selected to represent PCBs.

• For PAHs, the contaminants were selected based on the results of cluster analysis

performed on PAH mass fractions. Clustering is the partitioning of a dataset into

subsets, or “clusters,” where the data in each subset share some common trait. The

PAH cluster analysis yielded three different clusters (Figure 3-3). The two

independent PAHs selected from two of the clusters as contaminants with unique

sources or fate and transport processes consist of Benzo(a)pyrene (from the green

group in Figure 3-3) and Fluoranthene (from the blue group in Figure 3-3). The

third cluster was not included because it contained mostly 2- and 3-ring PAH

compounds, which likely have significant dissolved phase concentrations and may


not be conservative particle tracers. Note that the EMB model focuses on particle-

bound contaminants.

• For pesticides, the selected compounds were limited by data availability and

difference in analytical techniques. In the TSI 1995 data set only Total DDx7

compounds were reported for the DDT group of compounds. In Newark Bay, only

dichlorodiphenyldichloroethylene (DDE) was consistently detected in the

sediments. Therefore, 4,4’-dichlorodiphenyldichloroethylene (4,4’-DDE) was

selected for the EMB model. In addition, gamma-Chlordane was selected because

dated sediment cores indicated that there has been little or no change in sediment

concentration over time and thus provides a good check on the model (See Data

Evaluation Report No. 3 in Appendix A).

• For the metals, cluster analysis was used to separate them into four different

clusters (Figure 3-4). Four metals were selected (chromium, copper, lead and

mercury), one from each cluster, as contaminants with unique sources or fate and

transport processes.

In addition to the above contaminants, the contaminant normalizers iron and TOC were

included to account for variability in particle size and organic carbon content of the

sediment. These normalizers helped to reduce the variability in the concentrations of

sediments and suspended solids (see Data Evaluation Report No. 4 in Appendix A).

The EMB model was designed to solve simultaneous mass balance equations for various

parameters by optimization. Thirteen parameters (eleven contaminants plus iron and

TOC) were directly used in the model for optimization (Table 3-1).

In addition to the list of 13 optimized parameters, another nine parameters were selected

for further EMB model evaluation (Table 3-2). This additional EMB model evaluation

was done by: 1) using model-optimized solids contributions to predict the concentrations

of the nine additional parameters in recently-deposited sediment of the Lower Passaic

7 Total DDx refers the sum of the 4,4’-dichlorodiphenyldichloroethane (4-4’-DDD), 4,4’-dichlorodiphenyldichloroethylene (4,4’-DDE) and 4,4’- Dichlorodiphenyltrichloroethane (4,4’-DDT) concentrations in a sample.


River, and 2) comparing the predicted concentrations to the observed values for these

parameters.

3.4 Best Estimate Scenario

The sources and the receptor used in the EMB model are shown in Figure 3-1. For

completeness, a brief description of each source and the receptor is provided here along

with their respective concentrations used for the model parameters. The concentrations of

these parameters represent the best estimates for the various sources/receptor; the

application of these concentrations in the EMB model is called the best estimate scenario.

• Resuspension of the FFS Study Area legacy sediments was represented by the

average surface concentrations from the 1995 TSI dataset (i.e., 0-6 inches surface

sediment from RM1 to RM7). The average contaminant concentrations for the

resuspension signature are summarized in Table 3-3.

• Newark Bay was characterized by a northern and southern region. Average

contaminant concentrations for these regions are shown in Table 3-4; however, the

Newark Bay end member is represented by the northern region in the base case

simulation given its proximity to the Lower Passaic River. The surface sediment (0-6

inch) samples used to delineate the Newark Bay end member were from the 2005

Phase 1 and 2007 Phase 2 RI study by TSI. Only surface sediments (0-6 inches) at

depositional locations in the channel were considered (see Data Evaluation Report

No. 2 in Appendix A for discussion).

• The Upper Passaic River was characterized by four Be-7-bearing surface sediment

samples (only two of these were analyzed for organic contaminants), four Be-7-

bearing dated sediment core tops, and the suspended solids from two sediment traps.

These samples were collected between 2005 and 2008. The average contaminant

concentrations for the Upper Passaic River are summarized in Table 3-5.

• Tributary concentrations were based on averages from several recently-deposited

surface sediment samples and sediment trap samples obtained during the 2007/2008

sampling event. The average contaminant concentrations for the tributaries are

summarized in Table 3-6. Water column suspended sediment samples were removed


from the population before the calculation of the statistics because these water

column samples represent a snap-shot at the time of collection (a few hours), and may

not be representative of average conditions. Indeed, several of them were reported to

have unusually high concentrations of many of the contaminants (possibly reflecting

rain event-driven peaks in contaminant concentrations). The exception is Second

River, where sediment and water column suspended sediment samples were used.

These water column suspended sediment samples did not show the variability

observed in other surface water samples and there were not enough sediment samples

to calculate meaningful statistics from sediment alone.

• The CSO and SWO data were based on water column suspended sediment samples

taken at the outfalls of several CSO and SWO locations (Table 3-7). The SWO

samples were determined not to be representative of the contribution of SWOs to the

contaminant loads in the river and they were not used in the base case model

simulation (see Section 3.2 for a discussion of the data quality from the SWO

samples).

• The recently-deposited Lower Passaic River surface sediments are the receptor in the

model. They were characterized by recently-deposited sediments, including core tops

from the 2005 high resolution cores and Be-7 surface sediment samples from the

2007/2008 sampling event. Data Evaluation Report No. 4 in Appendix A shows that

most contaminants have relatively constant iron-normalized concentrations from

RM2 to RM12, but these ratios often vary at the two ends of the study area. For this

reason, only data for samples between RM2 and RM12 were used in the model. The

average concentrations are listed in Table 3-8.

3.5 Sensitivity Analysis of SWO Concentrations and Solids Constraint

The base estimate scenario described above used the best estimates of the concentrations

for the various parameters for the sources/receptor in the EMB model. However, because

the SWOs were not sampled at a location above the influence of the Lower Passaic River,

the data from the SWO samples were compromised by the intrusion of Lower Passaic

River sediments into the SWOs. Therefore, the contributions from Second River and the

SWOs were combined in the model and the contaminant characteristics of both were


based on samples taken in Second River only (see Section 3.2 for a discussion of SWO

data quality). The Second River was deemed to be representative of SWO discharges into

the Passaic River, because the Second River drains a highly-urbanized watershed that is

fed primarily by storm water collection systems. To assess the impact of this premise on

the best estimate solution, a model scenario was conducted that separated the SWO from

the Second River, with the SWO contaminant profile represented by the average of the

compromised SWO data. The solids and contaminant contributions obtained from this

sensitivity scenario were compared with the corresponding results from the best estimate

solution.

The second sensitivity analysis performed was to assess the impact of the solids

constraint on the best estimate solution. The solids constraint states that the sum of the

solids fractions from the various sources in the objective function (Equation 3-4) equals

one. Because of differences in the particle size distribution from the various sources, the

sum of the solids fractions may not necessarily be equal to one. This constraint was tested

in a sensitivity analysis and the model solution was compared to the results for the best

estimate scenario.

3.6 Monte Carlo Analysis of Uncertainty and Variability in Contaminant

Concentrations

The best estimate scenario described above used the best estimates of the concentrations

for the various parameters for the sources and receptor in the EMB model; however to

account for uncertainties and variability in source and receptor compositions, a Monte

Carlo sampling approach was used to develop 10,000 iterations of the input parameters

and the EMB model was optimized for each set of input parameters (i.e., 10,000

optimized solutions were obtained). The objective of the Monte Carlo analysis was to

develop confidence bounds in the EMB model-estimated solids balance and contaminant

fate and transport deduced from the solids balance by accounting for uncertainties in

source and receptor composition, and in the spatial variability in parameter

concentrations. Detailed description of the Monte Carlo simulation approach is given in


Attachment A. In brief, the Monte Carlo simulation approach was used to develop the

10,000 iterations of the input parameters as follows:

• For the external sources and the receptor concentrations, a bounded normal

distribution defined by the mean, standard deviation, minimum, and maximum of

each parameter in each source term and receptor was used to perform the Monte

Carlo simulation.

• For resuspension, a bootstrap8 method was used to simulate the 10,000 iterations

of the contaminant concentrations in resuspended sediment since the 1995 TSI

data are neither normal nor log-normal.

• The correlations amongst the parameters for each source and receptor were

examined to verify that the 10,000 iterations of parameter profiles represent the

contaminant inter-dependencies.

As stated previously, 10,000 iterations were used to create 10,000 optimized model

estimates of the solids concentrations. Those 10,000 estimates of the solids contributions

were used to developed confidence levels of the solids contribution and the sources

parameter contributions to the Lower Passaic River.

3.7 Model Performance Evaluation

Model-optimized receptor concentrations for the 13 optimized parameters and model-

predicted receptor concentrations for the nine additional contaminants were evaluated for

the best estimate scenario using a statistical indicator called the normalized mean error

(NME). The NME is defined as the difference between the predicted and the observed,

normalized to the observed receptor concentration for each parameter (Equation 3-5):

measured

measuredel

CCCNME −

= mod Equation 3-5

8 Bootstrap is a powerful Monte Carlo method that re-samples the original sample set with replacement to generate a distribution of sample's statistics. It is a non-parametric method.


where,

Cmodel = Parameter-specific concentration estimated

by the model

Cmeasured = Parameter-specific concentrations measured

in the Lower Passaic River

The NME expresses the bias in model predictions and observations, and gives an

indication of overestimation (NME >0) or underestimation (NME <0) for each

contaminant.

3.8 Model Limitation

Receptor models are inferential in nature, meaning that they infer the contributions from

different sources based on multivariate measurements collected at the receptor site.

Because the models infer rather than predict, they cannot be used directly to estimate

future changes in the system under certain conditions. For example, while the model

indicates that a fraction of the Lower Passaic River bottom sediments is composed of

Newark Bay sediments, the model cannot predict how the Newark Bay contribution will

change after the Newark Bay channel is deepened.


4 MODELING RESULTS

This chapter discusses the EMB model solids balance results and the fate and transport of

contaminants deduced from the solids balance in the Lower Passaic River. The EMB

model calculations were performed for the best estimate scenario and a Monte Carlo

analysis was included to assess the uncertainty in model estimates. In addition, two

sensitivity analysis scenarios were performed to assess the sensitivity of the model result

to the inclusion of compromised SWO sample results and to the use of tributary solids

constraints included in the model. The model results for the best estimate scenario form

the basis for the solids balance and contaminant fate and transport in the river. The results

of the Monte Carlo analysis were used to account for uncertainties and spatial variability

on the best estimate model results, and these uncertainties were expressed as confidence

levels on the best estimate solution. The Monte Carlo analysis also provided a median

estimate based on the 10,000 iterations which was also compared to the best estimate

scenario.

4.1 EMB Model Solids Balance Results: Best Estimate Scenario and

Uncertainty

Thirteen parameters [Table 3-1; copper, chromium, mercury, lead, gamma-Chlordane,

4,4’-DDE, 2,3,7,8 TCDD, Total TCDD, Total PCB, benzo(a)pyrene, fluoranthene, iron,

and TOC] were optimized in the model to determine the solids balance. The model was

then used to predict the receptor concentrations for the remaining nine parameters in

Table 3-2 to evaluate its performance (see Section 4.3 below). Note that while iron and

TOC are not contaminants, they are generally important in the transport of fine particles

and associated contaminants. In particular, the inclusion of iron and TOC in the EMB

model is an indirect means of normalizing the various source terms to their fine-grained

sediment content. Therefore, the EMB model focuses on those sediments that contain and

transport the majority of the contaminant burden.


Uncertainties in the model solution were developed from the Monte Carlo analysis based

on confidence intervals (5th and 95th percentiles) of the 10,000 optimized solutions. The

results of the EMB model optimization of the 10,000 Monte Carlo iterations are

presented later in this discussion as box and whisker plots which depict the median

solution plus the 5th, 25th, 75th and 95th percentiles of the 10,000 iterations. The best

estimate solution was also added to the plots.

The EMB model solids balance results are shown in Figure 4-1 based on best estimates of

the concentrations of contaminants. The best estimate solution indicates that resuspended

solids account for about 48 percent of the total solids in recently-deposited (Be-7-

bearing) sediments in the Lower Passaic River. Newark Bay and the Upper Passaic River

account for about 14 percent and 32 percent, respectively, of the solids delivered to the

Lower Passaic River. The tributaries, CSO and SWO together contribute about 6 percent

of the solids. Uncertainties in these solids fraction estimates derived from the Monte

Carlo iterations (Figure 4-2) indicate that resuspension accounted for about 28 to 65

percent, Upper Passaic River accounted for about 13 to 49 percent, Newark Bay

accounted for less than 1 to 44 percent, and all the other sources together contribute

between 2 and less than 12 percent. The relatively high contribution of solids from

resuspension translates to a high resuspension contribution (33 percent or higher) of the

contaminant burden (Table 4-1) in recently-deposited (Be-7-bearing) sediments.

4.2 EMB Model Contaminant Fate and Transport

The EMB model solids balance results presented above (Section 4.1) lead to further

discussion of the fate and transport of contaminants in the Lower Passaic River. The fate

and transport discussions are based on the mass balance outputs showing the distribution

of the contaminant flux among the sources and a comparison of average contaminant

concentrations used to characterize each source. This section is divided into two sub-

sections: (1) fate and transport of parameters examined and optimized in the EMB model,

and (2) inferred fate and transport of additional parameters. The results for the best

estimate scenario and the associated Monte Carlo-based uncertainty are presented for the

contaminants examined in the EMB model. Only the best estimate scenario is presented


for fate and transport of additional contaminants, since these contaminants were not part

of the Monte Carlo analysis.

4.2.1 Fate and Transport of Contaminants Optimized in the EMB Model

2,3,7,8-TCDD and Total TCDD Mass Balances

The upper panel of Figure 4-3a presents a box and whisker plot of the 2,3,7,8-TCDD

concentration for each source with a solid line (marked “Target Concentration9”)

representing the average 2,3,7,8-TCDD concentration in recently-deposited (Be-7-

bearing) sediments in the Lower Passaic River (from RM2 to RM12). The first striking

feature is that external sources alone (the Upper Passaic River, the tributaries, the

CSO/SWOs, and Newark Bay) cannot explain the measured 2,3,7,8-TCDD concentration

in the river. Note that the 2,3,7,8-TCDD concentrations from the upland external sources

(the Upper Passaic River, the tributaries, the CSO/SWOs) are approximately two orders

of magnitude less than the measured concentration in the recently-deposited (Be-7-

bearing) surface sediments. Northern Newark Bay is approximately one order of

magnitude lower, likely due to the impacts of the Lower Passaic River on this water

body. Consequently, another source of 2,3,7,8-TCDD is necessary to achieve a closed

contaminant mass balance. The only other source that could explain the target

concentrations in the Lower Passaic River is the resuspension of legacy sediments. The

mass balance calculated for 2,3,7,8-TCDD, shown on Figure 4-3a bottom panel and

Figure 4-3b, indicates that resuspension accounts for about 87 to 100 percent, with a best

estimate of 97 percent of the 2,3,7,8-TCDD observed in recently-deposited sediments in

the Lower Passaic River.

Similar results were observed for Total TCDD (Figure 4-4a,b); however, for Total

TCDD, the relative difference between the measured concentration in the Lower Passaic

River and the concentrations in the external sources is less than the corresponding

difference observed for 2,3,7,8-TCDD. The Total TCDD concentration difference among

the upland external sources is only about one order of magnitude, as opposed to two

9 “Target concentration” represents the average contaminant concentration in recently-deposited (Be-7-bearing) sediments in the Lower Passaic River between RM2 and RM12.


orders of magnitude for 2,3,7,8-TCDD. Even though the variation is much less, sediment

resuspension is still necessary to achieve a closed contaminant mass balance. While all

external sources account for about 1 to 28 percent with a best estimate of 8 percent of the

Total TCDD mass balance, sediment resuspension accounts for 76 to 97 percent with a

best estimate of ~ 92 percent of the contaminant mass.

The Newark Bay contribution to the Lower Passaic River dioxin contaminant burden

ranges from less than 1 percent to 13 percent for 2,3,7,8-TCDD, with a best estimate of 3

percent. For Total TCDD, Newark Bay contribution ranges from less than 1 percent to 21

percent, with a best estimate of 5 percent. The concentrations of these contaminants in the

river surface sediments are greater than the reported concentrations in the bay. These

results indicate that the Lower Passaic River is a source of contamination to the bay.

Total PCB Mass Balance

The fate and transport of Total PCBs is influenced by sediment resuspension (Figure 4-

5a, b). The Total PCB concentration in Newark Bay is about two times lower than the

Lower Passaic River concentration and the Upper Passaic River concentration is about

three times lower than the Lower Passaic River concentration. These concentration

patterns in the source signatures indicate a dominant resuspension contribution of Total

PCBs to the Lower Passaic River with a best estimate of about 81 percent and a range of

59 to 90 percent. Upper Passaic River is the most important external source of total PCB

contamination to the Lower Passaic, with a best estimate of 11 percent (range of 4 to 22

percent of the overall mass), while the Newark Bay contributes about 7 percent (range of

less than 1 to 25 percent) of the overall mass.

PAH Mass Balance

In the model, benzo[a]pyrene (Figure 4-6a, b) and fluoranthene (Figure 4-7a, b) represent

the PAH contaminant compounds directly optimized by the model. For both of these

compounds, the average PAH concentration in the Upper Passaic River is higher

(approximately 1.5 times) than the Lower Passaic River average PAH concentration. The

tributaries, the CSOs/SWOs and the 1995 surface sediment concentrations are


comparable to the measured PAH concentration in the 2005-2007 Be-7-bearing

sediments in the Lower Passaic River. The Newark Bay average PAH concentration is

about two times smaller than the Lower Passaic River concentration. Because the average

PAH concentration in the Upper Passaic River is higher than the target concentration

(Lower Passaic River) and the other sources have similar PAH concentration, the

contribution of the Upper Passaic PAH to the Lower Passaic PAH contamination is larger

than any other compounds used in the mass balance. The Upper Passaic River

contribution ranges from 27 to 70 percent for benzo[a]pyrene, with a best estimate of 53

percent and for fluoranthene from 24 to 64 percent with a best estimate of 47 percent.

Resuspension of the historical inventory accounts for about 39 percent (range of 17 to 58

percent) of the PAH contaminant burden of the Lower Passaic River. Newark Bay’s PAH

contribution range from less than 1 percent to 30 percent, with a best estimate of

approximately 6 percent.

Although higher PAH concentrations were observed in the tributaries and CSOs,

comparable to observations in the Upper Passaic River, the relatively small solids

contributions from the tributaries and CSOs limits their combined contribution to less

than 17 percent.

Pesticides Mass Balance

The average 4,4’-DDE concentration in the Upper Passaic River is roughly four times

lower than the measured concentration in the Lower Passaic River (Figure 4-8a, b). The

4,4’-DDE concentration in Newark Bay is slightly lower than the concentration in the

Lower Passaic River. The average 4,4’-DDE concentration of the 1995 surface sediment

source is only slightly higher than the 2005-2007 Be-7-bearing sediments in the Lower

Passaic River (approximately 30 percent higher). While the Second and Third River 4,4’-

DDE concentration overlaps with measured 4,4’-DDE in the 2005-2007 Be-7-bearing

sediments in the Lower Passaic River, the limited solids load from these tributaries

cannot account for the 4,4’-DDE mass in the river. The resuspension of the historical

inventory contributes between 52 to 88 percent of the 4,4’-DDE mass in the Lower

Passaic River, with a best estimate of 78 percent. Newark Bay contributes about 8 percent


of 4,4’-DDE mass (range of less than 1 to 34 percent) and the Upper Passaic River

contributes about 10 percent (range of 4 to 21 percent). The combined contribution from

tributaries SWOs and CSOs range from 1 to 10 percent, with a best estimate of about 4

percent.

The gamma-Chlordane concentrations in the tributaries are about two to four times higher

than that of the measured gamma-Chlordane in the 2005-2007 Be-7-bearing sediments in

the Lower Passaic River (Figure 4-9a, b). Notably, both the Upper Passaic and the 1995

0-6 inch surface sediment have a lower gamma-Chlordane concentration compared to the

measured gamma-Chlordane in the 2005-2007 Be-7-bearing sediments in the Lower

Passaic (about 30 percent lower). Since the average gamma-Chlordane concentrations in

the tributaries, SWOs and CSOs are higher, the combined contribution to the Lower

Passaic River from these sources ranges from 4 to 25 percent, with a best estimate of 13

percent. While this fraction is relatively small, it is the amount needed to raise the

gamma-Chlordane concentration in Be-7-bearing sediments above the concentrations

observed in sediments from the Upper Passaic River and Newark Bay, as well as in

resuspended sediments. The Upper Passaic contributes about 32 percent of the gamma-

Chlordane contamination to the Lower Passaic. The resuspension of the historical

sediment inventory accounts for about 32 to 70 percent of the gamma-Chlordane

contamination in the Lower Passaic River, with a best estimate of 52 percent. Newark

Bay contribution ranges from less than 1 to 19 percent of the gamma-Chlordane

contamination to the Lower Passaic River, with a best estimate of approximately 3

percent. The gamma-Chlordane contributions from the Upper Passaic River, Newark Bay

and resuspension remain significant because of the relatively large mass of solids

contributed by these sources.

Metals Mass Balance

Similar to 4,4’-DDE, the fate and transport of copper, chromium, mercury, and lead in

the Lower Passaic River is dominated by sediment resuspension. In the case of copper

(Figure 4-10a top panel), higher concentrations relative to the target concentration were

observed in the Second River/SWOs and the CSOs, as well as the resuspension source.


Copper concentrations in the other sources were less than the average target

concentration. Given that the solids contribution from the Second River and CSOs are

insignificant relative to the resuspension contribution, a mass balance for copper would

need a significant resuspension contribution to explain the high target concentration. This

is confirmed by the model-estimated copper budget (Figure 4-10a bottom panel and

Figure 4-10b), which shows that resuspension accounts for 72 percent of the contaminant

burden (range of 45 to 85 percent), while Newark Bay, the Upper Passaic and CSOs

account for 12 percent (range of less than 1 to 40 percent), 14 percent (range of 5 to 25

percent) and 1 percent (range of less than 1 to 6 percent), respectively.

The relative concentrations of chromium (Figure 4-11a top panel) and mercury (Figure 4-

12a top panel) in the various sources show higher or comparable average concentrations

in Newark Bay and resuspension sources and lower concentrations for other sources,

relative to the target concentrations. A mass balance for these metals can only be

obtained by a large resuspension contribution to explain the target concentrations. This

observation is confirmed by the best estimate and Monte Carlo mass balance results for

chromium (Figure 4-11a bottom panel) and mercury (Figure 4-12a bottom panel).

Resuspension of sediment accounts for approximately 74 percent of chromium and

approximately 75 percent of mercury, both with a range between 44 and 88 percent. Both

chromium and mercury also have similar contributions from Newark Bay and the Upper

Passaic River, with respective values of 15 and 10 percent for chromium and 14 and 11

percent for mercury.

Average concentrations of lead from the various sources are shown in the top panel of

Figure 4-13a. Lead concentrations are higher in the Second River/SWOs, CSOs and the

resuspension source, relative to the average target concentration in the Lower Passaic

River. Because the solids contribution from the Second River/SWO and CSOs are

relatively small, a significant resuspension input is needed to explain the observed target

concentration. The mass balance calculated for lead (Figure 4-13b, bottom panel)

indicates a best estimate of 71 percent resuspension contribution to the overall lead

burden in recently-deposited (Be-7-bearing) sediments, with an uncertainty range of 48 to


83 percent. The best estimates of the lead contributions from Newark Bay and the Upper

Passaic River are about 7 and 19 percent, respectively, while the tributaries, SWO and

CSOs contribute a combined 3 percent.

Iron and TOC Balance

Average source concentrations for iron indicate higher iron concentrations and, by

association, higher fractions of fine particles in Newark Bay and the Lower Passaic River

sediments relative to the other sources (Figure 4-14a top panel). The resulting mass

balance (Figure 4-14a bottom panel) indicates between 29 to 72 percent, with a best

estimate of 54 percent of the iron in Be-7-bearing sediments originating from

resuspension. The iron contribution from Newark Bay ranges from less than 1 to 52

percent, with a best estimate of approximately 18 percent. The Upper Passaic River

contributes a best estimate of 24 percent (range of 9 to 43 percent) of the iron burden to

the target area.

Unlike iron, which indicates an appreciable Newark bay contribution, TOC in Newark

Bay is low relative to other sources (Figure 4-15a, top panel). The TOC mass balance

indicates a best estimate resuspension contribution of 72 percent to the TOC burden in

the Lower Passaic River, with a range of 48 to 83 percent (Figure 4-15a bottom panel).

4.2.2 Inferred Fate and Transport Model for contaminants

Two contaminant mass balances could not be fully quantified in the EMB model due to

data gaps/limitations and the degree of particle affinity of any given contaminant. For

example, dieldrin was generally not detected in the Phase 1 Newark Bay dataset, thus

only the Phase 2 data with some detected values were used and an inferred best estimate

mass balance was developed for dieldrin.

Furthermore, because Low Molecular Weight (LMW) PAHs may be affected by

dissolved phase concentrations as well as other contaminant degradation processes, they

were not explicitly included in the EMB model. However, the best estimate EMB model

solids balance was used to calculate a mass balance for phenanthrene, used as a surrogate


for LMW PAH compounds. A summary of the inferred best estimate mass balances is as

follows:

• The inferred best estimate mass balance (Figure 4-16) for dieldrin compares with that

estimated for 4,4,’-DDE, with resuspension, Newark Bay, and the Upper Passaic

River accounting for 73 percent, 7 percent, and 12 percent of the target burden,

respectively.

• Phenanthrene was characterized at each source (the Upper Passaic River, tributaries,

CSOs/SWOs, Newark Bay, and resuspension) and was used as a surrogate to

represent LMW PAH. Unlike the other PAH compounds, benzo[a]pyrene (Figure 4-

6) and fluoranthene (Figure 4-7), which indicate slightly higher contributions from

Upper Passaic River relative to resuspension, the phenanthrene mass balance (Figure

4-17) indicates about a 47 percent and 42 percent resuspension and Upper Passaic

River contributions, respectively. Newark Bay contributes about 2 percent of the

phenanthrene in the Lower Passaic River, comparable to Saddle River with 4 percent.

4.3 Evaluation of EMB Model Performance

Model-calculated receptor concentrations for all parameters were evaluated through a

statistical indicator referred to as the NME, which was defined in Section 3.7. Note that

the NME expresses the bias in model predictions and observations, and gives an

indication of overestimation (NME >0) or underestimation (NME <0) for each parameter.

Estimates of the NME indicate that the best estimate EMB model optimization resulted in

predicted recently-deposited concentrations in the Lower Passaic River for the 13

parameters within 25 percent of the observed values, with the exception of gamma-

Chlordane, which is under-predicted by 38 percent (see red columns in Figure 4-18).

Evaluation of model performance for the remaining nine parameters (see blue columns in

Figure 4-18) also shows very good fits, with an NME within 25 percent for most

parameters, with the exception of indeno(1,2,3-cd)pyrene, which is under-predicted by 28

percent.


4.4 Sensitivity Analysis

Uncertainties in source and receptor compositions and spatial variability in parameter

concentrations were incorporated into the model solution through the Monte Carlo

analysis described above. This section presents the result of additional analysis performed

to assess the impact of compromised SWO contaminant concentrations (Table 4-3) and

the model solids constraint (Table 4-4). The sensitivity results are discussed below.

Impact of Stormwater Data

The impact of compromised SWO data was evaluated by performing a model simulation

using separate source compositions for Second River and the SWOs (the SWO

contribution was separated from the Second River, with the SWO contaminant profile

represented by the average of the compromised SWO data). The results of this scenario

were not significantly different relative to the best estimate scenario (Tables 4-2 and 4-3,

respectively). As expected, the use of the SWO data as an individual source did not affect

the model calculations, most likely, because the SWOs, like the other tributaries, are

minor solids contributors.

Impact of Model Solids Constraint

The best estimate solution, which was simulated using a solids constraint which required

that the total solids fraction should be one, was redone without this constraint. There were

slight differences between the best estimate solution (Table 4-2) and the relaxed solids

constraint scenario (Table 4-4). In general, the differences were within the uncertainty

estimated by the best estimate scenario of the Monte Carlo analysis. When the solids

constraint was relaxed, the model predicted the solids fractions with an error of 8 percent,

a value within the variability inherent in the contaminant measurements. The agreement

between these two scenarios suggests that the contaminants profiles provide adequate

constraint on the mass balance, as well as a strong mathematical basis to track the

sediment types that are mostly associated with the contamination.


4.5 Assessment of the Cooperating Parties Group (CPG) 2008-2009 Data

In this mass balance model, the receptor concentration was defined by the contaminant

concentration measured in Be7-bearing surface sediment samples collected during the

USEPA 2005 and 2007 sampling events (representing the top 1 inch of sediment). The

sources of fine-grained solids to the river were the Upper Passaic River, tributaries, CSOs

and SWOs (sampled in the 2007-2008 sampling events), and Newark Bay (sampled as

part of the Phase I and II field investigation of Newark Bay, TSI 2007, 2008). The source

characteristics for resuspension of Lower Passaic River sediments were represented by

the surface sediment (0-6 inch) concentrations from the 1995 TSI dataset.

In 2008, the CPG collected low resolution cores; however, the samples did not

characterize recently-deposited sediments (i.e., Be-7-bearing). In the main stem of the

Passaic River, surface sediment concentrations between the 1995 TSI and the 2008 CPG

were compared in Data Evaluation Report No. 4 in Appendix A, and the results indicate

that median surface sediment contaminant concentrations have not changed much

through this period.

For completeness of the EMB model, the 0 to 6-inch surface sediment concentrations

reported by the CPG for the tributaries and Upper Passaic River were compared to the

data used in the models. In the mass balance, each source term was defined by a Monte

Carlo simulation to generate a bounded-normal distribution of possible contaminant

concentrations. Table 4-5 provides a comparison of contaminant concentrations reported

for the CPG 2008 surface sediments along with the Monte Carlo simulation range that

was used to characterize each source term in the mass balance. For comparison purposes,

the actual USEPA samples that were used in the Monte Carlo simulation are also

provided. For all four external sources (Saddle River, Second River, Third River, and

Upper Passaic River), the CPG 2008 surface sediment data generates average

contaminant concentrations that fall outside the Monte Carlo simulation range. The CPG

2008 Dundee Dam data (Upper Passaic River) were generally higher than the Monte

Carlo simulation range, while the CPG 2008 tributary data were lower than the range. As


discussed above (and presented in Table 4-5), this difference is likely associated with

sampling depths that represent different physical/contaminant regimes. Above Dundee

Dam, 0 to 6 inches of silty sediment is likely characterizing deeper legacy sediments, so

higher sediment concentrations are expected; whereas on the sandy tributaries, a 0 to 6-

inch sample is likely capturing the underlying sand, which reduces the overall sample

concentration. For the two tributaries with higher 2,3,7,8-TCDD concentrations than used

in the model analysis, these stations are likely to be impacted by tidal transport of Lower

Passaic River sediments into the tributaries during low flow periods. Consequently, the

existing contaminant mass balance does not need any modification since the CPG 2008

sampling event was not designed to characterize the source term or solids transported

from these sources to the river, and in some cases may be impacted by solids from the

Lower Passaic River itself.


5 FORECASTING CONTAMINANT CONCENTRATIONS

The goal of this chapter is to integrate several of the analyses and observations described

in the previous chapters to develop a basis to forecast future contaminant concentrations

in surface sediments. The forecasting formulation represents the river section between

RM2 and RM12 as a single system consisting of 1) a water column where mixing of

particles from external sources and resuspension occurs; and 2) a mixed-layer surface

sediment bed to which particle deposition from the water column occurs. The rationale

for using this representation of the river section from RM2 to RM12 is based on

observations of recently-deposited sediments which show little longitudinal variation in

median concentrations (see Data Evaluation Report No. 4 in Appendix A). There are

concentration gradients in the recently-deposited sediments at either end of this river

section which represent the mixing zones with Upper Passaic River (from RM12 to

RM17) and Newark Bay (from RM0 to RM2). Note however that this observation related

to recently-deposited sediments does not suggest that surface sediments show little

variability. As shown in Data Evaluation Report No. 4 in Appendix A, surface sediment

concentrations of the various contaminants vary by several orders of magnitude. The

variability in surface sediment concentration (as well as other sources of variability) was

accounted for stochastically by a Monte Carlo simulation approach for the forecasting

analysis, providing an estimate of future contaminant concentrations in the river bed.

The forecasting analysis integrated the relative solids and contaminant contribution

results from the EMB model, the observed surface sediment concentrations, current

contaminant compositions of external sources, and historical trends of sediment

contamination from dated sediment cores, as discussed below. Similar to the EMB model

analysis, a Monte Carlo analysis consisting of 10,000 iterations was performed to

quantify uncertainties in contaminant forecasts for the single system. These forecasts can

be used to estimate the future contaminant concentrations in the 0-6 -inch surface

sediment layer, the interval which corresponds to the bioactive sediment layer. The

ability to predict future exposure point concentrations in this horizon is important for risk

assessments and to evaluate the FFS remedial alternatives.


This chapter is divided into two major sections: “Development of Concentration Half

Times10 for the Excess Passaic River Sediment Burden” and “Forecasts of Sediment

Concentrations for FFS Remedial Alternatives.” The first section describes the results of

an analysis which uses the dated sediment cores to examine the decline in contaminant

concentrations over time. As part of this analysis, baseline contaminant levels in the

external upland sources are subtracted from the Lower Passaic River sediment

concentrations observed in the dated cores, yielding the component of the annual

contaminant burden that is due to loads “internal” to the Lower Passaic River. This is also

referred to as the “excess sediment burden.” Nearly all internal contaminant loads present

in the Lower Passaic River can be attributed to legacy sediment resuspension. It is

essential to quantify the internal burden since it represents the contamination that would

be controlled by a remedial action performed on the Lower Passaic River. Using the

observed decline of this burden over time, the rate of concentration decrease can be

described by a first-order exponential decay curve with an estimated half time based on

the rate of the natural recovery process (i.e., the time it takes for the concentrations in

depositing sediment to decline by half as a result of these processes). Once the half times

for the different contaminants were determined, regression analysis was used to compare

the contaminant-specific half times and determine a single half time for the excess

sediment burden in the Lower Passaic River.

In the second section, the half times developed from the dated sediment core

chronologies for the internal resuspension load were used, together with the results of the

EMB model, to forecast future concentrations in the bioactive sediment layer. The

forecast calculations were made for the best estimate scenario using average contaminant

concentrations and other model inputs, and uncertainties were quantified by Monte Carlo

analysis. These forecasts are made for the FFS Remedial Alternatives, including: No

10 The use of the term “half time” in this sense is not to imply decay or destruction of 2,3,7,8-TCDD over time, akin to the decay of a radionuclide. Rather, the term here is used to express a rate for the decline of 2,3,7,8-TCDD concentrations in the solids accumulating at each coring location. Specifically, the half time is the time required for the 2,3,7,8-TCDD concentration to decline to half of its current value. The processes that affect the decline are multifold, including many of the fluxes and processes that occur in an urban estuary. The “half time” expression is just a means to encompass these processes and note their net effect on concentration through time.


Action (Alternative 1), Deep Dredging with Backfill (Alternative 2), Capping with

Dredging for Flooding and Navigation (Alternative 3), and Focused Capping with

Dredging for Flooding (Alternative 4). Detailed description of these alternatives is

provided in Chapter 4 of the FFS Report.

5.1 Development of Concentration Half Times for the Excess Passaic River

Sediment Burden

The dated sediment core profiles for the Lower Passaic River and the Upper Passaic

River at Dundee Dam describe the chronologies of contaminant concentrations in the

sediment. By careful selection of the coring locations and radionuclide dating of the

sediment layers, the time-dependence of contaminant concentrations can be discerned. As

discussed in Data Evaluation Report No. 3 in Appendix A these sediment records are a

proxy for contaminant concentrations on suspended matter in the water column at the

time of deposition. Because tidal mixing integrates suspended matter and the associated

contaminant loads over distances of several miles, each dated sediment core records the

relative intensity of loads as they are deposited on the river bottom in its vicinity.

To the extent that the core records yield regular variations over time (e.g., a steady

decline in contaminant concentrations from depth to the surface), the trends in the core

chronologies can be extrapolated and used as a basis to estimate future conditions in the

absence of remediation. Essentially, the rate of contaminant concentration decline

documented by the core implies a rate of recovery for the river’s sediments in the absence

of any marked changes in loads or processes. In most instances, these loads and processes

(e.g., the integration of a large watershed area) are difficult to change or redirect without

major intervention.

The dated sediment cores document the impacts of internal and external loads to annual

deposition, the equivalent of the annual Be-7-bearing sediment deposits. In order to

forecast future impacts of legacy sediment-related loads, it is first necessary to

distinguish the component of the water column-based sediment record (i.e., the dated


sediment cores) that is due to legacy sediment resuspension from that which is due to

external loads.

The EMB model, presented in Chapters 3 and 4, used available field data to represent the

contaminant sources associated with sediment flux from each of the major tributaries, the

CSOs/SWOs, Newark Bay, and the Upper Passaic River. The EMB model also used the

1995 surface sediment data to estimate the conditions of the internal contaminant source:

the resuspension of legacy sediments. After balancing the loads from all the sources with

the known conditions of recently-deposited (Be-7-bearing) sediment, the EMB model

produced a set of fractions describing the solids contribution from each of the sources.

These fractions can be used to separate the loads measured in the dated sediment cores

into the contributions from each source, which is necessary to identify which portion of

the contaminant load will be controlled by remediation. The underlying premise of this

approach is that the relative solids contributions from each of the solids sources to annual

deposition as recorded in the cores has remained constant over for the historical period

examined. This is for the period 1980 to 2005 as described below.

5.1.1 Natural Recovery Processes Occurring in the Lower Passaic River

Natural recovery processes are likely occurring in the Lower Passaic River and impacting

some contaminant concentrations over time. These trends are observed in the dated

sediment core profiles presented in Data Evaluation Report No. 3 in Appendix A, with

concentrations declining from the 1980s to 2005 for some contaminants. Table 5-1

summarizes the average 1980s concentrations for some of the contaminants and

compares these values to the average 2005 surface sediment concentrations.

The observed concentration decline may be due to multiple factors, including: natural

recovery processes (such as mixing and burial) and the elimination of direct discharges,

the combination of which curtailed the contaminant load over time. For this analysis,

only sediment samples dated from 1980 and later were included. Based on the dated

cores, this period was inferred to represent natural recovery-type reductions in


concentration more than reductions caused by the elimination of direct contaminant

discharges. These natural reductions are assumed to continue into the future.

5.1.2 Data Available for Estimating the Half Times and Developing the

Trajectories

Each of the sources and the data available to quantify the decline in concentrations over

the last three decades are described herein:

• Lower Passaic River – There are five high resolution sediment cores collected in

2005 that document the characteristics of fine-grained suspended solids over time,

representing 12 miles of the Lower Passaic River and providing a basis for an

analysis of sediment contaminant concentrations as they change through time (see

Data Evaluation Report No. 3 in Appendix A).

• Upper Passaic River – One of the high resolution cores taken from Dundee Lake in

2005 by scientists from Rensselaer Polytechnic Institute (RPI) contains sufficient data

to describe the trend in contaminant concentrations in the Upper Passaic River over

the last 30 years. This core is presented in Figures 3-1, 3-2, 3-5 and 3-6 in Data

Evaluation Report No. 2 in Appendix A. With the exception of PAHs and dieldrin,

the detected contaminants have had relatively constant concentrations in the

suspended sediments above Dundee Dam since about 1990. Previous to that time,

there were elevated contaminant concentrations in the core segments and, by

inference, in the contemporaneous suspended solids transported over the dam. PAHs

and dieldrin appeared to decline in concentration until about 1990, when the trend

reversed and concentrations began to increase again. In addition, data from several

core tops, surface sediment samples, and sediment traps collected in 2007-2008 were

used to estimate the current average concentration on suspended sediment (Be-7

bearing).

• Newark Bay – No known high resolution cores exist to quantify

depositional/contaminant chronologies in the Newark Bay sediments for the post-

1990 period. The surface sediment samples from 2005 indicate a spatial gradient for

many of the contaminants from south to north, but there is no information on

temporal change in sediment concentrations. The EMB model used the five northern


samples (see Table 3-4) to define the Newark Bay end member for use in the model.

The trajectory calculations also used these northern samples to calculate the Newark

Bay component.

• Tributaries – There are no temporal data available to determine the changes of

sediment contaminant concentrations over time in the Saddle River, Second River or

Third River.

• CSO/SWOs – There are no temporal data available to determine the changes of

contaminant concentrations over time in the releases from either the CSOs or the

SWOs.

To determine the temporal changes in the internal (resuspension) Lower Passaic River

contribution to the total contaminant load, each of the external sources was quantified and

subtracted from the total concentration, using the solids fractions obtained from the EMB

model results.

In the absence of information to the contrary, the tributaries and the CSO/SWO

components were assumed to have constant contaminant concentrations from the 1980s

to the end of the trajectory forecast. Since the EMB model found that the combined

sediment contributions from these sources were less than 5 percent of the entire sediment

load in the river, their contribution is small enough to warrant an assumption of this

nature without materially affecting the outcome of the trajectory forecasts.

The Upper Passaic River contribution was defined by a linear interpolation of

concentration versus time between each core segment from the RPI core (see Data

Evaluation Report No. 3 in Appendix A). After the last data point on that core (2005) and

through the end of the trajectory forecasts, the Upper Passaic component was assumed to

be constant. The constant concentration was an average of the core tops from two RPI

cores (one previously mentioned) and two Malcolm Pirnie cores collected early in 2007,

as well as a number of surface sediment and sediment trap samples from the 2007/2008

sampling program.


The remaining two components are the Lower Passaic River resuspension component and

the Newark Bay component. Given the close link between these two water bodies and the

lack of a Newark Bay core to separately track fine-grained suspended matter from the

Bay, there is no way to separate their rates of decline. As a result, both contributions are

assumed to decline at the same rate. Therefore, when calculating the rate of decline, the

portion of the total contaminant load remaining after subtraction of the Upper Passaic

River, tributary, and CSO/SWO loads is defined as the “excess load”.

The assumption that Newark Bay concentrations are declining at a similar rate to the

Lower Passaic River sediments is rational, and perhaps conservative, given that the

northern end of Newark Bay is the end member used in the EMB model. Data Evaluation

No. 2 in Appendix A discusses the evidence to support the premise that 70 percent of the

dioxin load in Newark Bay is derived from the Lower Passaic River. Since so much of

the sediment dioxin load in Newark Bay originates from the river, it is appropriate to

assume that the rates of concentration decline are similar. For other contaminants, the

Lower Passaic River’s contribution to Newark Bay is probably much less due to other

sources. However, in these instances, the Newark Bay concentrations at the southern end

of the Bay are used as the “base” for the Newark Bay contribution (i.e., northern Newark

Bay concentrations are not permitted to decline below this concentration). Thus for those

contaminants with a strong north-to-south gradient in Newark Bay (suggesting an

important Lower Passaic River contribution), the large difference between the ends of the

bay is allowed to decline at the rate observed for the Lower Passaic River. For those

contaminants with a shallow or no gradient, the concentration on Newark Bay solids

delivered to the Lower Passaic River remains essentially constant over time.

5.1.3 Calculating the Half Times

Figures 5-1 through 5-10 show the excess concentration, obtained by subtracting the

products of each upland source’s concentration (based on averages of available data) and

solids fraction (from the EMB model) from the concentrations assigned to each slice of

the five high-resolution cores from the Lower Passaic River. The resulting datasets were

fitted to a first-order exponential decay curve as described above. While the exact


mechanism(s) and rate of decline for this excess concentration are not specifically

known, the results clearly indicate a decline for most contaminants. The choice of a first

order decay process is consistent with the expected processes affecting most

contaminants in this system, specifically dispersion, bioturbation, diffusion, and

degradation. For each of these processes, the rate at which they occur is linearly

dependent on the contaminant concentrations (e.g., the higher the concentration, the

higher their rates of dispersion, diffusion, and degradation). The basic first-order decay

equation is presented as Equation 5-1 below.

tot eCC λ−= Equation 5-1

Where

Ct: excess sediment concentration at a given time

Co: excess sediment concentration at the initial time

λ: exponential decay parameter

t: time

Moreover, the exponential decay parameter (λ) is related to the half time (Equation 5-2),

or the estimated time for the contaminant concentration to decrease by half:

λ)2ln(

=halft Equation 5-2

Where

thalf : time estimated for the 1980 concentration to decline by half

The regression fits of the exponential regression lines, coefficient of determination of the

fits (R2), confidence interval of the half times, and the level of significance of the

regression are included in Figures 5-1 through 5-10. The fits of the exponential regression

lines were shown to be statistically significant (P<0.05) for all of the parameters except


gamma-Chlordane and the sum of High Molecular Weight (HMW) PAHs. For these two

contaminants, the confidence intervals, especially the upper interval, cannot be estimated.

Further, one contaminant, dieldrin, was shown to have an increasing, statistically

significant trend. An increasing trend cannot be explained by the geochemical constructs

inherent in this analysis, so dieldrin was not forecast.

Forecasting the future behavior of LMW PAHs was not performed due to the inability of

the EMB model to balance the contribution of the various sources to the Lower Passaic

River for these contaminants, which may indicate that their fate and transport is not

strictly tied to fine-grained sediments. However, the measured trend of LMW PAH

concentrations from the dated sediment cores, which is declining with a half time of

about 63 years (Figure 5-10) provides an indication of sediment recovery for LMW

PAHs.

The individual contaminant-specific half times and associated confidence intervals for all

contaminants are listed in Table 5-2. Note that these are not true half times for the

contamination in the Lower Passaic River; rather, they are half times for the portion of

the contamination that is attributable to resuspension in the Lower Passaic River and

input from Newark Bay (“excess concentration”). This is the only portion that is assumed

to be declining exponentially. Comparison of the confidence intervals of 2,3,7,8-TCDD,

total PCB, 4,4’-DDE, mercury, lead, copper, and gamma-Chlordane shows significant

overlap suggesting a common exponential decline for the excess sediment burden in the

Lower Passaic River. To estimate this common half time for the excess sediment burden,

a first-order regression model was developed incorporating the excess contaminant

concentrations for multiple contaminants and their estimated time of deposition in the

Lower Passaic River. Details of this regression analysis are described in Attachment B.

The results of the analysis indicate a common average half time of approximately 35

years for the excess sediment burden. The 95 percent confidence interval for this

common half time is from 27 to 48 years. Although only seven contaminants were

included in the model, this result also applies to other particle-reactive contaminants in

the Lower Passaic River that have a significant resuspension source term.


5.2 Forecasts of Sediment Concentrations for FFS Remedial Alternatives

This section describes the process to determine the best-estimate, post-remediation

contaminant concentrations and associated uncertainties in surface sediments for the

following four alternatives:

• No Action (Alternative 1)

• Deep Dredging with Backfill (Alternative 2)

• Capping with Dredging for Flooding and Navigation (Alternative 3)

• Focused Capping with Dredging for Flooding (Alternative 4)

For each alternative listed, post-remediation surface sediment concentrations were

forecast for the contaminants listed in Table 5-1. All of the remedial alternatives listed

above include the 200,000 cubic yards of sediment removed behind a coffer dam under

the Tierra Removal Phases 1 and 2 (see FFS Report Section 4.0 for more information).

However, this removal action was not explicitly included in the empirical trajectory

forecast for the following reasons:

• Surface sediment concentrations for 2,3,7,8-TCDD in the Tierra Removal Phase 1

and 2 areas compare to the range of values reported for other locations within the

FFS Study Area. This observation also applies to other contaminants. Thus

inclusion of these values does not affect the mean concentration estimates or the

associated statistics used in the model.

• Because the removal occurred within confinement, release of extremely high

concentration in the deeper sediment layers in Tierra Removal Phase 1 and 2

areas is not anticipated.

• The forecast trajectory model represented the surface sediments in the FFS Study

as a single system represented by the average. Excluding the post-dredging

anticipated concentrations of approximately zero in the spatially small Tierra

Removal Phase 1 and 2 areas does not affect the average sediment concentrations

used to represent the legacy sediments in the model or the variability in

concentrations represented in the Monte Carlo simulation.


5.2.1 Overview of Remedial Alternatives

Alternative 1 - No Action – Although the No Action Alternative involves no active

remedial technologies, natural recovery processes (e.g., mixing and burial) may be at

work to reduce contaminant concentrations in sediments over a period of interest.

Alternative 2 – Deep Dredging with Backfill – This active remedial alternative specifies

removal of the fine-grained sediments present in the FFS Study Area, bank to bank, by

dredging. The intent of Alternative 2 is to remove as much contaminated fine sediment as

practicable between RM0 and RM8.3. Dredging outside the Tierra Removal Phase 1 and

2 areas would begin in March 2018 and all activities will be completed in October 2028,

and backfill placement will be completed in July 2029. The release of sediments and

contaminants during the dredging process is not represented in the empirical forecast

model.

Alternative 3 – Capping with Dredging for Flooding and Navigation – This active

remedial alternative specifies a combination of dredging and capping, bank to bank, of

the fine-grained sediments present in the FFS Study Area. The intent of Alternative 3 is

to sequester the contaminated sediments under an engineered cap, while dredging enough

material to limit flooding that might be caused by the installation of a cap and to

accommodate current and projected future use of the federal navigation channel from

RM0.0 to RM2.2. Dredging outside the Tierra Removal Phase 1 and 2 areas would begin

in March 2018 and will be completed in November 2022. Backfill and cap placement

activities will be completed in December 2022. The release of sediments and

contaminants during the dredging processes is not represented in the empirical forecast

model.

Alternative 4 – Focused Capping with Dredging for Flooding – This active remedial

alternative specifies a combination of dredging and capping of discrete areas of fine-

grained sediments that add up to about one-third of the river bottom in the FFS Study

Area. Dredging outside the Tierra Removal Phase 1 and 2 areas would begin in March

2018 and all activities will be completed in February 2020. Final cap placement is


anticipated to be completed in March 2020. It was assumed that one-third of the area will

be remediated. The release of sediments and contaminants during the dredging processes

is not represented in the empirical forecast model.

5.2.2 Assumptions for the Active Remediation Alternatives

The methodology used to derive the trajectory forecasts are presented in detail in

Attachment C. In order to forecast contaminant concentrations in surface sediments after

active remediation (Alternatives 2 to 4), several assumptions were needed:

• From 1995 to the end of 2017, the concentration trends for the contaminants for all

alternatives will continue to follow an exponential decline based on the half time

values provided in Table 5-2 and estimated from contaminant histories obtained from

the high-resolution cores. Beyond 2017, only the No Action alternative will continue

this trend through the end of the trajectory analysis.

• For Alternatives 2, 3, and 4, a linear reduction in surface sediment concentration

during the implementation of the remedy is assumed between the value in 2017 and

the anticipated value at the end of the remedy. For Alternatives 2 and 3, surface

sediment concentrations declined linearly from the value in 2017 to zero in 2029 and

2022, respectively. For Alternative 4, surface sediment concentrations declined

linearly by one-third from the value in 2017 to 2020.

• After remediation is complete, the impact of any remedy on resuspension is

proportional to the fraction of fine-grained sediment area addressed by the remedy.

This premise is based on the observation that the majority of the contaminant burden

is associated with fine-grained sediments. Thus the reduction in the resuspension

contribution declines directly with the reduction in fine-grained sediment surface

area. Therefore, for Alternatives 2 and 3, remediation of sediments from RM0 to

RM8.3 results in a 75 percent reduction in the sediments and contaminants available

for resuspension (e.g., erosional silt areas) over the entire 17 miles of the Lower

Passaic River. The exception to this is 2,3,7,8-TCDD. Because 2,3,7,8-TCDD is not

found in the sediments above RM12 to an appreciable degree, the availability of

2,3,7,8-TCDD contaminated fine-grained sediment for resuspension is reduced by 88

percent (see the formula derivation in Attachment C).


• The average sedimentation rate (0.27 inches/year) remains unchanged by the remedy.

• Following remediation, surface sediment concentrations in unremediated areas will

continue to decline exponentially with the same half time values provided in Table 5-

2. This assumption does not account for any reductions in the flux of contaminants

carried into the unremediated areas after the remediation of the lower eight miles. It

also does not account for any remediation from RM8.3 to RM17.4 that might be

planned by the Cooperating Parties Group (CPG). This assumption would tend to

underestimate the benefits of remediating the lower eight miles.

• Initially, after remediation, the resuspension contribution from the capped or

backfilled areas will be zero, due to the designed resistance to erosion or the use of

sand as backfill. However, over time, it is anticipated that fine-grained sediments will

settle on the cap, recreating the current sediment texture (i.e., a fine-grained area that

is capped with coarse sand will, over time, become covered with fine-grained

sediments again). Thus, the volume resuspended from the remediated areas was

allowed to linearly increase each year until a full 6-inch biologically active layer has

been developed on top of the remediated area. Assuming 0.27 in/year of

sedimentation this is calculated to take 22 years. After the 22-year period,

resuspension from the remediated area will be at the same rate as for any other fine-

grained area in the river.

• No decline in concentrations in the Upper Passaic River, the tributaries, or the

CSO/SWOs will occur at any time in the future. Concentrations of Newark Bay

suspended matter delivered to the Lower Passaic River will reduce exponentially

(with the half times listed in Table 5-2) towards the level currently measured in the

Be-7 bearing sediment at the southern end of the bay. Thus, the surface sediment

concentrations in the Lower Passaic River will asymptotically approach the level of

contamination represented by the combination of these external sources.

The final premise listed above refers to a “floor” or baseline value for each trajectory.

Although the Newark Bay source is assumed to be dropping exponentially, it is not

reasonable to assume that it will reach undetectable levels for most contaminants within

the time period analyzed without remediation efforts. This analysis assumes that the


interactions between Newark Bay and the Lower Passaic River continue as they are now,

and that these interactions will not be impacted by changes to the system (e.g.,

maintenance dredging). The samples from the northern part of the bay were used as the

end members for the EMB model. For many contaminants, Newark Bay has an

increasing contaminant concentration trend from south to north towards the mouth of the

Passaic River. This indicates mixing of relatively cleaner southern Newark Bay

sediments with comparatively more contaminated sediments from the Lower Passaic

River. Thus, for the purposes of this analysis, the southern Newark Bay sediments were

assumed to represent the lowest concentration that Newark Bay sediments can achieve

without active remediation in the Bay.

Similarly, the concentration assigned to the resuspension component within the Lower

Passaic River was also assigned a “floor” value. Because the resuspended sediment is

comprised of sediments introduced from the sources as well as legacy contaminated

sediments, it is natural to assume that the concentration of the resuspended sediment will

not drop below the sum of the external source contaminant loads. The floor for the

contaminant concentrations on resuspended sediments was calculated as the sum of the

products of the constant sources (Upper Passaic River, tributaries, CSO/SWOs) and their

solids fractions and the product of the Newark Bay “floor” value and the Newark Bay

solids fraction.

This floor value was implemented into Equation 5-1 as follows:

( ) fefCC tot +−= −λ

Equation 5-3

Where:

Ct: sediment concentration at a given time

Co: sediment concentration at the initial time

f: “floor” value calculated as described above.

λ: exponential decay parameter derived in Section 5.1

t: Time


In this way, the concentration of surface sediments approaches f, instead of 0 as given by

Equation 5-1.

5.2.3 Calculation of the Best Estimate Trajectories

As described in the previous section, half times were calculated for each of the

contaminants of concern listed in Table 5-1 by subtracting out the portions of the total

concentration from the high resolution cores that were attributable to the Upper Passaic

River, the tributaries, or the CSOs and SWOs. The exponential line fit through the data,

then, represented the reduction in contaminant concentration due to natural recovery

processes occuring in the Lower Passaic River and Newark Bay.

Using the average concentrations for each of the sources and the half time calculated

from the high resolution cores, a natural recovery (no action) trajectory for annual

deposition (equivalent to the Be-7-bearing sediment deposited in a given year) was fit

through the concentrations for each source.

The EMB model was an optimization routine that systematically searched for the best set

of solids contribution fractions that would balance the mass of each contaminant from

each source. It was an over-constrained set of 13 equations with 7 unknowns. Because of

modeling assumptions, simplifications to the system, and variations or errors in data sets,

the optimization routine was not able to find a solution that perfectly balanced all 13

parameters. As explained in Chapter 4, the fit was good for most of the parameters. In the

case of gamma-Chlordane, the fit was not as good, resulting in an error of 38 percent.

Because of this lack of fit, the summation of the gamma-Chlordane contributions from

each of the sources did not match the data collected in the main stem of the river between

RM2 and RM12 during the 2007 sampling effort or the core tops from the 2005 high

resolution cores. For most of the contaminants, model prediction yielded a sufficiently

small NME that the contaminant trajectories passed through the concentration range

represented by the 2005-2007 data. For gamma-Chlordane however, the trajectory fell

significantly below the recent data points, and each source concentration was artificially


increased by the percent error from the EMB model to bring the trajectory up to match

the measured data.

As discussed above for the active remedial alternatives, treatment of the river bottom is

expected to prevent resuspension of contaminated sediment. However, these treatments

would also initially prevent resuspension of any sediment on the cap or backfill. Because

the cap or backfill will be composed of clean sand and, in the case of a cap, may be

armored in certain areas to prevent erosion due to high velocities, this assumption would

apply for the early part of the remedy’s design life. As time goes on, the river is expected

to deposit silt particles on the top of the cap or backfill. These particles are then available

for resuspension and mixing with other sources from the river. The amount of silt

available for resuspension within the remediated area would increase from zero at the

time of installation to a maximum level, at which point the resuspension from the

remediated area would match the resuspension from unremediated areas.

In the calculation of these trajectories, it was assumed that the rate of resuspension from

the remediated area would match that of unremediated areas once a 6-inch layer of

sediment had been re-deposited on top of the capped or backfilled area. This layer would

be biologically and physically active, and was assumed to be a vertically mixed surface

layer. At a sedimentation rate of 0.27 inches per year on average, it would take 22 years

for this layer to reach an average depth of 6 inches. For simplicity, the amount of

resuspension on the cap or backfill was assumed to increase linearly from 0 to the level of

the unremediated areas during this 22- year period. Since the remediation is assumed to

be completed in 2029 (Alternative 2), 2022 (Alternative 3) and 2020 (Alternative 4), the

newly re-deposited layer would be 6 inches thick by the years 2051, 2045 and 2042,

respectively.

Since the calculations allow the remediated areas to eventually contribute to the

resuspended sediment load in the river, it is necessary to estimate the concentration of the

contaminants in this resuspended sediment. This material is derived from the upper 6

inches of the capped or backfilled surface. The contaminant concentrations in this layer


initially would be zero, since the cap or backfill is assumed to be free of contaminants.

This concentration would not be added to the river in the first year since the material

would be designed to largely remain in place. As the remediated surface begins to

accumulate sediments from the water column, it becomes more subject to resuspension.

The concentration of contaminants in sediments resuspending from the remediated areas

was calculated assuming that the upper 6 inches represents a mix of newly-deposited

sediment and the cap or backfill material. Thus, in the second year, the upper 6 inches

would equal on average 5.73 inches of clean cap or backfill material and 0.27 inches of

deposited sediment. The contaminant concentration of this layer would be the volumetric

average of the concentration on the newly deposited 0.27 inches of sediment and the 5.73

inches of clean material. Continuing this process, in the third year the contaminant

concentration in the upper 6 inches would be a volumetric average of 5.73 inches of

mixed cap or backfill material (from the previous year) with a concentration equal to that

calculated the second year and 0.27 inches of newly-deposited material with a

concentration equal to the forecast Be-7-bearing surface sediment concentration from the

previous year. In each case, the two concentrations are mixed proportional to the

thicknesses mentioned. This composite concentration was then applied to any

resuspension occurring from the remediated areas for that year.

For the purposes of the risk analysis and for comparison of the remediation alternatives, it

is most appropriate to compare the effects of the remediation on the upper 6 inches of

sediment instead of simply the recently-deposited (Be-7-bearing) sediment. This is

because the upper 6 inches represents what is typically considered the biologically active

layer and thus represents the exposure point concentration for most biota. For the No

Action alternative, the mean sediment concentration over the upper six inches of

sediment can be estimated by averaging the annual sediment deposition for the previous

22 years as characterized by the dated sediment cores. When the upper 6-inch layer

concentration for each of the contaminants is estimated in this way for the year 1995, all

of the contaminant concentrations estimated from the dated sediment cores (except Total

PCBs) fall within two standard errors of the mean value from the TSI 1995 surface

sediment (0-6 in) dataset.


The 1995 TSI dataset does not report concentrations for all the PCB congeners, so the

Total PCB concentration was estimated as the sum of the Aroclors. This differs from

other samples in the 2005 and 2007/2008 sampling events, which report all of the

congeners, and the Total PCB as a sum of congeners. These two sums are not always

analogous. A regression analysis between Aroclor and congener-based total PCB

estimates using 2005 to 2010 data indicated that Total PCBs by Aroclor was biased low

by 25 percent (see Data Evaluation Report No 5 in Appendix A). Despite this difference,

the Total PCB concentrations estimated by the EMB model agreed to within 2 percent of

the measured data.

Assuming that the concentration trends observed for the past 25 years will continue into

the future, surface sediment concentrations from 2005 to 2059 were forecast for the

annual deposition (equivalent to the Be-7-bearing sediments deposited each year) for

each contaminant following Equation 5-4. These concentrations were then used to

estimate the exposure concentrations in the 0-6 inch biologically active layer of sediment

using the process outlined in Attachment B. The placement of sand material to construct

a sub-aqueous cap or to backfill after dredging is assumed to temporarily restrict fine-

grained sediment resuspension, as described above. Given the many unknowns associated

with remedial construction sequence for the various alternatives, no attempt has been

made to predict short-term consequences during implementation, such as resuspension

during dredging; these short-term impacts are incorporated into the mechanistic model

(Appendix B) and addressed in the FFS. During the remediation, a linear reduction in

surface sediment concentration is assumed between the value in 2017 and the anticipated

value at the end of the remedy. After completion of each remedy, for the purposes of

long-term forecasts, it has been assumed that all benefits of remediation are realized in

the final year of the typical remedial alternative construction period (2029, 2022, or 2020

for Alternatives 2, 3, and 4, respectively). The following equation integrates the spatial

extent of remediation and the processes affecting the 0 to 6 inch layer:


( )( )( ) ( ) RSP

ii

RSP

capRSPRSPREM fcb

fCfcb

bcCdbCfC

−−+

−−

+−−= ∑

11111

Equation 5-4

Where:

CREM: new recently-deposited (Be-7-bearing) sediment concentration following

remediation in 2029, 2022, or 2020 for Alternatives 2, 3, and 4, respectively.

CRSP: contaminant concentration for resuspending sediment (top six inch average)

Ccap: contaminant concentration on top of the cap or backfill (top six inch average)

Ci: contaminant concentration on incoming sediment from source, i

i: subscript representing all sediment sources other than resuspension

fRSP: fraction of solids associated with resuspension

fi: fraction of solids associated with source, i

b: fraction of fine-grained sediment area impacted by placement of cap or backfill

material

c: ratio of resuspension occurring in remediated areas per unit area to that in

unremediated areas

d: adjustment for fine-grained sediment area not impacted by remediation and not

contaminated with dioxin (this value is zero for all contaminants other than

dioxin.)

5.2.4 Calculation of the Trajectory Uncertainties

The best estimate trajectory calculations described above used the best estimates, or

averages of all the concentrations and other inputs needed, for the trajectory forecast

calculations. The uncertainties in best estimate trajectories were determined by Monte

Carlo simulation for 2,3,7,8-TCDD, Total PCB, 4,4’-DDE, mercury, lead, copper, and

gamma-Chlordane. Detailed description of the Monte Carlo analysis, which was

composed of 10,000 iterations of randomly-generated input values to the trajectory

calculation, is provided in Attachment A. The 10,000 Monte Carlo-generated inputs to

the contaminant forecasts calculations include:

• Randomly generated contaminant inputs to the EMB model;

• Optimized EMB model solids balance output;


• Randomly-generated half times using the confidence bounds and a normal

distribution assumption;

• Randomly-generated sedimentation rates developed using bootstrap analysis of

the difference between the 1989 and 2007 bathymetric surfaces; and

• Randomly-generated estimates of the depth of the sediment mixed layer between

10 and 20 cm.

Uncertainties in the contaminant forecasts developed from the Monte Carlo analysis were

based on the confidence interval (5th, 25th, 75th and 95th percentiles) of the 10,000

optimized solutions.

5.2.5 Best Estimate Trajectory for the No Action Alternative (Alternative 1)

Assuming that the concentration trends observed for the past 25 years will continue into

the future, concentrations in the 0-6 inch surface sediment layer from 1995 to 2059 were

forecast for each contaminant following the procedure outlined in Attachment B. This

procedure assumes that the exponential decline noted in the combination of Lower

Passaic River (internal resuspension) source and the Newark Bay source continues at the

same rate noted for the last three decades. It also assumes that the “constant” sources

(Upper Passaic River, tributaries, and CSO/SWOs) remain constant for the period of

model simulation. The best estimate No Action Alternative concentration forecast is

calculated using Equation 5-3 for Newark Bay and the resuspension term. All other

sources are added in as constants. The best estimate No Action forecasts are presented in

part A of Figures 5-11 through 5-17 (e.g., Figure 5-11A) and Figure 5-18. Table 5-3

presents the percent reduction in forecast concentration in 2059 relative to 2017 level

before the active remedies are to be implemented.

Part A of Figures 5-11 through 5-17 and Figure 5-18 also show the mean concentration

reported in the 1995 TSI surface sediment dataset (red point). The error bars indicate two

standard errors above and below the mean. The estimated concentration for the 0-6 inch

biologically active zone is within this envelope in every case (see discussion on gamma-

Chlordane adjustment above).


These results indicate that 2,3,7,8-TCDD surface sediment concentrations are forecast to

decline by 59 percent (from 0.5 to 0.2 µg/kg) from 2017 to 2059 under No Action. For

the other contaminants, the decline is 50 percent or less (Table 5-3a for percentages and

Table 5-3b for the concentrations). The relatively larger decline in 2,3,7,8-TCDD

compared to other contaminants is because the external sources (e.g., Upper Passaic

River, tributaries, CSOs/SWOs) are not significant contributors of 2,3,7,8-TCDD to the

sediments of the Lower Passaic River. Contaminants like gamma-Chlordane and PAHs,

which have seen no appreciable decline in the last 25 years, are not predicted to decline at

all over the time period of the forecast. The smaller reductions seen for metals, DDE, and

PCBs indicate that significant sources of these contaminants exist in the Upper Passaic

River or Newark Bay. Although significant concentrations may exist in the tributaries or

the CSO/SWOs, their small solids contribution prevents them contributing a large mass

of hydrophobic contaminant to the Lower Passaic River.

5.2.6 Best Estimate Trajectory for Alternatives 2 to 4.

The other remediation alternatives involve dredging and capping/backfill described in

Section 5.2.1. The forecast concentrations in the 0-6 inch layer are shown as orange lines

for Alternative 2, green lines for Alternative 3, and purple lines for Alternative 4 on part

A of Figures 5-11 through 5-17 and Figure 5-18. The percent reductions in forecast

concentrations in 2059 relative to 2017 for each Alternative are tabulated in Table 5-3.

Overall, the trajectory results indicate that Alternative 4, the focused capping remedy,

provides much less reduction in future concentrations compared to Alternatives 2 and 3.

About a 90 percent reduction in 2,3,7,8-TCDD surface sediment concentrations was

estimated by 2059 for Alternatives 2 and 3 (from 0.5 to approximately 0.04 µg/kg), as

compared to about 70 percent for Alternative 4. The increase in surface sediment

concentrations after the implementation of the remedy shown in the figures follows from

infilling of sediments over the remediated areas from the external sources as well as

sediments upriver of the FFS Study Area.


5.2.7 Trajectory Uncertainties

Uncertainty in trajectory forecasts for contaminant concentrations in the bioactive layer

included using the uncertainty analysis developed for the EMB model combined with

additional variability in the distributions developed for the remaining parameters used in

the forecast model (e.g., “excess” contaminant half times, mixed layer thickness,

sediment deposition rate). Parts B, C, and D of Figures 5-11 through 5-17 present the

uncertainty in trajectory forecasts as confidence bounds associated with best estimate,

predicted future contaminant concentrations. The most significant finding from the

uncertainty analysis is that by 2059 there will be no statistical significant difference

between No Action (Alternative 1) and Alternative 4. In addition, Alternatives 2 and 3

are not different from each other by 2059. However, there is a statistically significant

difference between Alternatives 2 and 3 versus Alternatives 1 and 4.

5.3 Incorporation of the CPG 2008-2009 Data

The CPG 2008 and 2009 datasets were not available at the time that the Mass Balance

Forecast Model analysis was performed. The CPG 2008 and 2009 datasets provide an

opportunity to confirm the forecast Lower Passaic River surface sediment concentrations

that were presented in the previous sections. In Parts B, C, and D of Figures 5-11 through

5-17, the mean and two standards errors of the measured CPG 2008 and CPG 2009

surface sediment concentrations were added to the Monte Carlo trajectory results. For the

contaminants examined, the simulated 2008, and 2009 distributions of the mean surface

sediment concentrations fall within the uncertainty of the measurements, except for the

simulated 2009 mercury concentration distribution (which plots on the border of the

uncertainty range). This agreement between the model forecast and the new CPG data

provides a rough validation of the model and the original forecast.


6 SUMMARY

This appendix describes the EMB modeling analysis developed to support the FFS of the

lower eight miles of the Lower Passaic River. It consists of an EMB model designed to

characterize the fate and transport of contaminants in the Lower Passaic River, as well as

a semi-empirical model used to forecast the concentrations of contaminants in Lower

Passaic River surface sediment for the FFS remedial alternatives. The summary of the

observations are provided below.

6.1 Summary of EMB Results

• Resuspension of legacy sediments represented the single largest contributor of

solids to recently-deposited sediments, accounting for 28 to 65 percent of the

recent deposition with a best estimate of 48 percent. The Upper Passaic River

accounted for about 13 to 49 percent of recently-deposited solids, with a best

estimate of 32 percent. Newark Bay accounted for less than 1 to 44 percent, with

a best estimate of 14 percent. All the other sources together contribute between 2

and less than 12 percent.

• The mass balance calculated for 2,3,7,8-TCDD, indicates that resuspension

accounts for about 87 to 100 percent, with a best estimate of 97 percent of the

2,3,7,8-TCDD observed in recently-deposited sediments in the Lower Passaic

River.

• Resuspension contribution of Total PCBs to the Lower Passaic River ranges from

59 to 90 percent with a best estimate of about 81 percent. Upper Passaic River is

the most important external source of Total PCB contamination to the Lower

Passaic, with a best estimate of 11 percent (range of 4 to 22 percent of the overall

mass), while Newark Bay contributes about 7 percent (range of less than 1 to 25

percent) of the overall mass.

• External inputs of PAHs are very important to the PAH mass balance for the

system. The Upper Passaic River contribution for benzo[a]pyrene was estimated

as 53 percent (ranging from 27 to 70 percent) and for fluoranthene the estimate

was 47 percent (ranging from 24 to 64 percent). Resuspension of the historical


inventory accounts for about 39 percent (ranging from 17 to 58 percent) of the

PAH contaminant burden of the Lower Passaic River. Newark Bay’s PAH

contribution ranges from less than 1 percent to 30 percent, with a best estimate of

approximately 6 percent. Although higher PAH concentrations were observed in

the tributaries and CSOs, comparable to concentrations in the Upper Passaic

River, the relatively small solids contributions from the tributaries and CSOs

limits their combined contribution to less than 17 percent.

• For 4,4’-DDE, resuspension of the historical inventory contributes between 52 to

88 percent of the mass in recently-deposited sediments, with a best estimate of 78

percent. Newark Bay contributes about 8 percent of 4,4’-DDE mass (ranging from

less than 1 to 34 percent) and the Upper Passaic River contributes about 10

percent (ranging from 4 to 21 percent). The combined contribution from

tributaries SWOs and CSOs range from 1 to 10 percent, with a best estimate of

about 4 percent.

• Similar to 4,4’-DDE, the fate and transport of copper, chromium, mercury, and

lead in the Lower Passaic River is dominated by sediment resuspension.

• Overall, the EMB Model identifies the sediments of the Lower Passaic River as

an important source of all contaminants of concern and the single most important

source of 2,3,7,8-TCDD to the entire Lower Passaic River.

6.2 Summary of Contaminant Forecast Results

• Surface sediment concentrations of 2,3,7,8-TCDD are forecast to decline by 59

percent from 2017 to 2059 under No Action. For the other contaminants, the

decline is less than 48 percent. The relatively larger decline in 2,3,7,8-TCDD

compared to other contaminants is because the external sources (e.g., Upper

Passaic River, tributaries, CSOs/SWOs) are not significant contributors of 2,3,7,8-

TCDD to the sediments of the Lower Passaic River.

• The forecasts for Alternatives 2 and 3 show a significant decline over time for

most contaminant concentrations when compared to No Action. About a 90

percent reduction in 2,3,7,8-TCDD surface sediment concentrations was


estimated by 2059 for Alternatives 2 and 3. Sediment 2,3,7,8-TCDD

concentrations, which originate largely from an internal source (resuspension), are

most dramatically reduced by Alternatives 2 and 3. Contaminants such as PAHs,

which have a significant external source, are impacted immediately upon

remediation, but the improvement wanes as contaminated sediments from external

sources are deposited on top of the remediated area and subsequently may

resuspend and continue mixing with the river’s solids load.

• The surface sediment contaminant concentrations after implementation of

Alternatives 2 and 3 are not statistically different from each other by 2059 for the

contaminants forecast.

• The trajectory results indicate that by 2059 there will be no statistically significant

difference between No Action (Alternative 1) and the focused remedy

(Alternative 4). Alternative 4 does not provide the sustained reduction in future

concentrations estimated for Alternatives 2 and 3.


7 ACRONYMS

2,3,7,8-TCDD 2,3,7,8-Tetrachlorodibenzo-p-dioxin

4,4’-DDE 4,4’-dichlorodiphenyldichloroethylene

Be-7 Beryllium-7

CMB Chemical Mass Balance

CPG Cooperating Parties Group

CSM Conceptual Site Model

CSO Combined Sewer Overflow

DDE Dichlorodiphenyldichloroethylene

DDT Dichlorodiphenyltrichloroethane

EMB Empirical Mass Balance

FFS Focused Feasibility Study

HMW High Molecular Weight

LMW Low Molecular Weight

µg/kg micrograms per kilogram

mg/kg milligram per kilogram

NME Normalized Mean Error

PAH Polycyclic Aromatic Hydrocarbon

PCB Polychlorinated Biphenyl

PCDD/F Polychlorobenzodioxin/furan

ppb parts per billion

R2 coefficient of determination of regression line fits

RI Remedial Investigation

RM River Mile

RPI Rensselaer Polytechnic Institute

SWO Stormwater Outfall

TOC Total Organic Carbon

Total DDx the sum of the 4,4’-dichlorodiphenyldichloroethane (4-4’-DDD),

4,4’-dichlorodiphenyldichloroethylene (4,4’-DDE) and 4,4’-

Dichlorodiphenyltrichloroethane (4,4’-DDT) concentrations


Total TCDD Total Tetrachlorodibenzodioxin

TSI Tierra Solutions, Inc.

USEPA United State Environmental Protection Agency


8 REFERENCES

Bzdusek, P.A., E.R. Christensen, A Li, and Q Zou. 2004. “Source Apportionment of

Sediment PAHs in Lake Calumet, Chicago: Application of Factor Analysis with

Nonnegative Constraints”. Environmental Science Technology. 38(1): 97-103.

Chaky, D.A. 2003. “Polychlorinated Biphenyls, Polychlorinated Dibenzo-p-Dioxins and

Furans in the New York Metropolitan Area; Interpreting Atmospheric Deposition and

Sediment Chronologies.” PhD Thesis, Rensselaer Polytechnic Institute (Troy, NY).

August 2003.

Imamoglu, I., K. Li and E.R. Christensen. 2002. “Modeling Polychlorinated Biphenyl

Congener Patterns and Dechlorination in Dated Sediments from the Ashtabula River,

Ohio, USA.” Environmental Toxicology and Chemistry. 21(11): 2283–2291.

Johnson, G.W., W.M. Jarman, C.E. Bacon, J.A. Davis, R. Ehrlich and R.W. Risebrough.

2000. “Resolving Polychlorinated Biphenyl Source Fingerprints in Suspended Particulate

Matter of San Francisco Bay”. Environmental Science and Technology. 34(4):552-559.

Ogura, I., M. Gamo, S. Masunaga and J. Nakanishi. 2005. “Quantitative Identification of

Sources of Dioxin-Like Polychlorinated Biphenyls in Sediments by a Factor Analysis

Model and a Chemical Mass Balance Model Combined with Monte Carlo Techniques”.

Environmental Toxicology and Chemistry. 24(2): 277–285.

Soonthornnonda, P. and E.R. Christensen. 2008. “Source Apportionment of Pollutants

and Flows of Combined Sewer Wastewater”. Water Research, Volume 42, Issues 8-9,

pp1989-1998.

Su, M., E.R. Christensen, J.F. Karls, S. Kosuru, I. Imamoglu. 2000. “Apportionment of

Polycyclic Aromatic Hydrocarbon Sources in Lower Fox River, USA, Sediments by a


Chemical Mass Balance Model.” Environmental Toxicology and Chemistry. 19(6):

1481–1490.

Watson, J.G., et al., 2004. “Protocol for Applying and Validating the CMB Model for

PM2.5 and VOC.” Prepared for the Office of Air Quality Planning & Standards,

Emissions, Monitoring & Analysis Division. EPA-451/R-04-001. Available online:

http://www.epa.gov/scram001/models/receptor/CMB_Protocol.pdf. Last accessed:

December 30, 2013.


http://www.epa.gov/scram001/models/receptor/CMB_Protocol.pdf


TABLES

Appendix C: Mass Balance Modeling Analysis 2014

Lower Eight Miles of the Lower Passaic River

Table 3-1: EMB Model Optimization Parameters

Chemical Class Chemical Name

Metals Chromium

Copper

Lead

Mercury

PCDD/F 2,3,7,8-TCDD

Total TCDD

Pesticides 4,4’-DDE

gamma-Chlordane

PAH Benzo[a]pyrene

Fluoranthene

PCB Congeners and Co-Elutions Total PCB

Other Parameters Total Organic Carbon (TOC)

Iron

Table 3-2: Additional Parameters Used for EMB Model Evaluation

Chemical Class Chemical Name

Metals Arsenic

Cadmium

Cobalt

Nickel

Zinc

PAH Compounds Benz[a]anthracene

Chrysene

Indeno[1,2,3-cd]pyrene

Pyrene



Table 3-3: Average Resuspension (Lower Passaic River) Concentrations for Selected Contaminants (0-6

inch Surface Sediment Samples)

Analyte Resuspension (Lower Passaic River) Concentrations a

Arsenic (mg/kg) 11

Cadmium (mg/kg) 5.1

Chromium (mg/kg) 150

Cobalt (mg/kg) 11

Copper (mg/kg) 230

Lead (mg/kg) 330

Mercury (mg/kg) 3.3

Nickel (mg/kg) 45

Zinc (mg/kg) 560

gamma-Chlordane (μg/kg) 26

4,4’-DDE (μg/kg) 66

2,3,7,8-TCDD (ng/kg) 810

Total TCDD (ng/kg) 960

Total PCB (ug/kg) 2,100

Benz[a]anthracene (mg/kg) 2.5

Benzo[a]pyrene (mg/kg) 2.4

Chrysene (mg/kg) 3.1

Fluoranthene (mg/kg) 5.2

Indeno[1,2,3-cd]pyrene (mg/kg) 0.9

Pyrene (mg/kg) 5.5

Iron (mg/kg) 25,000

TOC (%) 10

Concentrations rounded to two significant figures.

Note:

a: Data Source: 1995 TSI 0-6 inch surface sediment samples RM1 to RM7 were used for average

resuspension calculation.



Table 3-4: Newark Bay Northern End and Southern End Average Concentrations for Selected

Contaminants

Analyte Average Northern

Concentration b

Average Southern

Concentration c

Arsenic (mg/kg) 11 10

Cadmium (mg/kg) 1.5 0.63

Chromium (mg/kg) 110 67

Cobalt (mg/kg) 10 10

Copper (mg/kg) 130 82

Lead (mg/kg) 125 77

Mercury (mg/kg) 2.2 0.93

Nickel (mg/kg) 35 33

Zinc (mg/kg) 250 160

gamma-Chlordane (μg/kg)a 4.8 3.8

4,4’-DDE (μg/kg) 25 14

2,3,7,8-TCDD (ng/kg) 86 23

Total TCDD (ng/kg) 180 65

Total PCB (ug/kg) 580 260

Benz[a]anthracene (mg/kg) 1.5 0.44

Benzo[a]pyrene (mg/kg) 1.8 0.47

Chrysene (mg/kg) 1.7 0.44

Fluoranthene (mg/kg) 2.4 0.60

Indeno[1,2,3-cd]pyrene (mg/kg) 0.8 0.32

Pyrene (mg/kg) 2.7 0.65

Iron (mg/kg) 30,000 30,000

TOC (%) 2.6 1.7

Concentrations rounded to two significant numbers.

Note:

a. Newark Bay samples from the 2005 sampling event were reported non-detect for gamma-Chlordane.

The value used here is from Phase 2 Dataset, 2007.

b. The samples used to delineate the northern Newark Bay end member were NB01SED46, NB01SED47,

NB01SED52, NB01SED52 (dup), NB01SED55 and NB01SED61 from Phase 1 dataset and NB02SED078,

NB02SED094, NB02SED104, NB02SED106, and NB02SED107 from Phase 2 dataset.

c. The five samples used to delineate the southern Newark Bay end member were NB01SED017,

NB01SED021, NB01SED024, NB01SED030 and NB01SED031 from Phase 1 dataset.



Table 3-5: Upper Passaic River Recently-Deposited Surface Sediment Concentrations for Selected

Contaminants

Analyte Upper Passaic River Concentrations a

Arsenic (mg/kg) 2.9

Cadmium (mg/kg) 1.5

Chromium (mg/kg) 31

Cobalt (mg/kg) 8.8

Copper (mg/kg) 63

Lead (mg/kg) 130

Mercury (mg/kg) 0.72

Nickel (mg/kg) 19

Zinc (mg/kg) 290


4,4’-DDE (μg/kg) 13

2,3,7,8-TCDD (ng/kg) 1.9


Total PCB (ug/kg) 420






Pyrene (mg/kg) 9.1

Iron (mg/kg) 16,000

TOC (%) 3.7


Note:

a: Samples from 2008 USEPA suspended-phase high flow storm sampling event were used to calculate the

average concentrations. Only recently deposited surface sediment samples were used in the above

calculation. Two water column suspended matter samples LPRP-LVCG-DDL-000004 and LPRP-LVCG-

DDL-000006 were not used. The samples used to delineate the Upper Passaic River were LPRP-SCSH-

DDL-000018, LPRP-SCSH-DDL-000068, LPRP-SCSH-DDL-000143, LPRP-SCSH-DDL-000153, LPRP-

SCSH-PSR-001607, LPRP-SCSH-PSR-001602, LPRP-SCSH-PSR-001604, LPRP-SCSH-PSR-001590,

LPRP-SCSH-PSR-001663, LPRP-SCSH-PSR-001579, and LPRP-SCSH-PSR-001589.



Table 3-6: Tributary Average Concentrations for Selected Contaminants a

Analyte Saddle River b

Third River c

Second River d

Arsenic (mg/kg) 3.6 5.7 3.4

Cadmium (mg/kg) 0.41 1.4 0.75

Chromium (mg/kg) 20 35 25

Cobalt (mg/kg) 4.4 5.7 4.9

Copper (mg/kg) 43 68 42

Lead (mg/kg) 57 150 170

Mercury (mg/kg) 0.1 0.48 0.26

Nickel (mg/kg) 10 20 21

Zinc (mg/kg) 150 260 220

gamma-Chlordane (μg/kg) 55 77 27

4,4’-DDE (μg/kg) 19 46 26

2,3,7,8-TCDD (ng/kg) 2.9 2 0.9

Total TCDD (ng/kg) 25 24 11

Total PCB (ug/kg) 370 400 100

Benz[a]anthracene (mg/kg) 2.8 3.4 2.92

Benzo[a]pyrene (mg/kg) 3.7 4.3 3.3

Chrysene (mg/kg) 4.6 5.6 4.2

Fluoranthene (mg/kg) 8.7 9.5 8.4

Indeno[1,2,3-cd]pyrene (mg/kg) 2.8 3.3 2.5

Pyrene (mg/kg) 7.3 8 7

Iron (mg/kg) 11,000 14,000 14,000

TOC (%) 4.1 5.5 4.5


Note:

a: Samples from 2008 USEPA suspended-phase high flow storm sampling event were used to calculate the

average concentrations. Only recently deposited surface sediment samples were used in the above

calculation.

b: The samples used to delineate the average Saddle River concentrations were LPRP-SCSH-SDR-000001,

LPRP-SCSH-SDR-000005, LPRP-SCSH-SDR-000006, LPRP-SCSH-SDR-000007, LPRP-SCSH-SDR-

000003, and LPRP-SCSH-SDR-000004. Samples were from 2008 USEPA suspended-phase high flow

storm sampling event.

c: The samples used to delineate the average Third River concentrations were LPRP-SCSH-THR-000001,

LPRP-SCSH-THR-000002, LPRP-SCSH-THR-000003, and LPRP-SCSH-THR-000006.

d: The samples used to delineate the average Second River concentrations were LPRP-SCSH-SCR-

000001, LPRP-SCSH-SCR-000004, LPRP-SCSH-SCR-000005, and LPRP-SCSH-SCR-000006. Samples

were from 2008 USEPA suspended-phase high flow storm sampling event.



Table 3-7: Average CSO and SWO Concentrations for Selected Contaminants

Analyte Average CSO

Concentrations a

Average SWO

Concentrations

Arsenic (mg/kg) 6.6 15

Cadmium (mg/kg) 2.1 1.9

Chromium (mg/kg) 68 100

Cobalt (mg/kg) 8.3 16

Copper (mg/kg) 310 260

Lead (mg/kg) 390 350

Mercury (mg/kg) 0.99 0.75

Nickel (mg/kg) 49 62

Zinc (mg/kg) 850 820

gamma-Chlordane (μg/kg) 30 120

4,4’-DDE (μg/kg) 25 60

2,3,7,8-TCDD (ng/kg) 4.2 20

Total TCDD (ng/kg) 73 123

Total PCB (ug/kg) 940 400

Benz[a]anthracene (mg/kg) 1.9 6.6

Benzo[a]pyrene (mg/kg) 2.2 9.7

Chrysene (mg/kg) 3.8 25

Fluoranthene (mg/kg) 5.7 39

Indeno[1,2,3-cd]pyrene (mg/kg) 2.1 9.2

Pyrene (mg/kg) 5.6 32

Iron (mg/kg) 22,000 42,000

TOC (%) 30 19


Note:

a: The samples used to delineate the average CSO concentrations were LPRP-LVCG-PSR-000399, LPRP-

LVCG-PSR-000400, LPRP-LVCG-PSR-000401, LPRP-LVCG-PSR-000402, LPRP-LVCG-PSR-000403,

LPRP-LVCG-PSR-000404, LPRP-LVCG-PSR-000405, LPRP-LVCG-PSR-000406, LPRP-LVCG-PSR-

000407, LPRP-LVCG-PSR-000423, LPRP-LVCG-PSR-000424, LPRP-LVCG-PSR-000433, LPRP-

LVCG-PSR-000434, and LPRP-LVCG-PSR-000435. Samples were from 2008 USEPA suspended-phase

high flow storm sampling event.

b: The samples used to delineate the average SWO concentrations were LPRP-LVCG-PSR-000409,


000413, LPRP-LVCG-PSR-000414, LPRP-LVCG-PSR-000416, LPRP-LVCG-PSR-000417, LPRP-

LVCG-PSR-000418, LPRP-LVCG-PSR-000419, LPRP-LVCG-PSR-000420, LPRP-LVCG-PSR-000421,


000428, LPRP-LVCG-PSR-000429, and LPRP-LVCG-PSR-000432. Samples were from 2008 USEPA

suspended-phase high flow storm sampling event.



Table 3-8: Average Lower Passaic River Recently-Deposited (Be-7 Bearing) Sediment Concentrations

for Selected Contaminants

Analyte Average Main stem (RM2 – RM12) Concentration a

Arsenic (mg/kg) 8.0

Cadmium (mg/kg) 3.6

Chromium (mg/kg) 110

Cobalt (mg/kg) 8.6

Copper (mg/kg) 160

Lead (mg/kg) 210

Mercury (mg/kg) 1.9

Nickel (mg/kg) 32

Zinc (mg/kg) 490


4,4’-DDE (μg/kg) 52

2,3,7,8-TCDD (ng/kg) 370


Total PCB (ug/kg) 1,200






Pyrene (mg/kg) 5.8

Iron (mg/kg) 26,000

TOC (%) 6.3


Note:

a: Data source: Be-7 bearing sediment samples from 2007-2008 Malcolm Pirnie Sediment Sampling

Program and 2005-2006 High Resolution Coring Program.



Table 4-1: Contaminant Burden Attributed to Resuspension

Analyte Percent of Contaminant Burden in Recently Deposited

Sediments Attributed to Resuspension and Confidence Interval

Chromium ≈ 74 (44 to 88)

Copper ≈ 72 (45 to 85)

Lead ≈ 71 (48 to 83)

Mercury ≈ 75 (43 to 88)

gamma-Chlordane ≈ 52 (32 to 70)

4,4’-DDE ≈ 78 (52 to 88)

2,3,7,8-TCDD ≈ 97 (87 to 100)

Total TCDD ≈ 92 (76 to 97)

Total PCB ≈ 81 (59 to 90)

Benzo[a]pyrene ≈ 33 (17 to 52)

Fluoranthene ≈ 40 (21 to 58)

Iron ≈ 54 (29 to 72)

TOC ≈ 72 (48 to 83)



Table 4-2: Summary of Solids Contribution Results for Best Estimate Scenarios

Solids and Analytes Resuspension Newark Bay Upper Passaic

River

Saddle

River

Second River / Storm

Water Outfall

Third

River

Combined Sewer

Overflow

Solids 48% 14% 32% 4% 1% 1% 0.5%

Copper 72% 12% 14% 1% 0.4% 0.3% 1%

Chromium 74% 15% 10% 1% 0.4% 0.3% 0.3%

Iron 54% 18% 24% 2% 1% 1% 0.5%

Mercury 75% 14% 11% 0.2% 0.2% 0.2% 0.2%

Lead 71% 7% 19% 1% 1% 0.5% 1%

gamma-Chlordane 52% 3% 32% 8% 2% 2% 1%

4,4’-DDE 78% 8% 10% 2% 1% 1% 0.3%

2,3,7,8-TCDD 97% 3% 0.1% 0.03% 0.003% 0.004% 0.005%

Total TCDD 92% 5% 3% 0.2% 0.03% 0.04% 0.07%

Total PCB 81% 7% 11% 1% 0.1% 0.3% 0.4%

Benzo(a)pyrene 33% 7% 53% 4% 2% 1% 0.3%

Fluoranthene 40% 5% 47% 5% 2% 1% 0.4%

TOC 72% 5% 17% 2% 1% 1% 2%



Table 4-3: Summary of Solids Contribution Results for SWO Sensitivity Scenarios


River

Saddle

River

Second

River

Third

River

Storm Water

Outfall

Combined

Sewer Overflow

Solids 47% 17% 31% 3% 0% 1% 1% 0%

Copper 69% 14% 13% 1% 0% 0% 2% 1%

Chromium 70% 18% 9% 1% 0% 0% 1% 0%

Iron 51% 22% 22% 2% 0% 0% 2% 0%

Mercury 72% 17% 10% 0% 0% 0% 0% 0%

Lead 69% 9% 18% 1% 0% 0% 2% 1%

gamma-Chlordane 50% 3% 30% 8% 0% 2% 7% 1%

4,4’-DDE 76% 10% 10% 2% 0% 1% 2% 0%

2,3,7,8-TCDD 96% 4% 0% 0% 0% 0% 0% 0%

Total TCDD 91% 6% 3% 0% 0% 0% 0% 0%

Total PCB 79% 8% 11% 1% 0% 0% 0% 0%

Benzo(a)pyrene 33% 8% 51% 4% 0% 1% 3% 0%

Fluoranthene 37% 6% 43% 5% 0% 1% 7% 0%

TOC 70% 6% 16% 2% 0% 1% 3% 2%



Table 4-4: Summary of Solids Contribution Results for Relaxed Solids Constraint Sensitivity Scenarios


River

Saddle

River

Second River / Storm

Water Outfall

Third

River

Combined Sewer

Overflow

Solids 42% 20% 32% 4% 2% 1% 0%

Copper 65% 18% 14% 1% 1% 0% 1%

Chromium 65% 23% 10% 1% 0% 0% 0%

Iron 46% 27% 23% 2% 1% 0% 0%

Mercury 67% 21% 11% 0% 0% 0% 0%

Lead 65% 12% 19% 1% 2% 1% 1%

gamma-Chlordane 48% 4% 33% 9% 2% 3% 1%

4,4’-DDE 72% 13% 11% 2% 1% 1% 0%

2,3,7,8-TCDD 95% 5% 0% 0% 0% 0% 0%

Total TCDD 89% 8% 3% 0% 0% 0% 0%

Total PCB 76% 10% 12% 1% 0% 0% 0%

Benzo(a)pyrene 30% 11% 53% 4% 2% 1% 0%

Fluoranthene 36% 8% 47% 5% 3% 1% 0%

TOC 67% 8% 18% 2% 1% 1% 2%


Page 1 of 2

Table 4‐5: Comparison of 2008 Tributary and Dundee Dam Measured Surface Sediment Concentrations with Monte Carlo Simulated Model Inputs

Second River Third River Saddle River

Contaminant

Units

Monte Carlo

Model Simulated Range of Possible

Concentrations

USEPA 2008 Surface Grab (0‐1 inch) and Sediment Trap

CPG 2008 Surface Results

(Measured 0 6

Monte Carlo


Concentrations

USEPA 2008 Surface Grab (0‐1

inch) and Sediment Trap

CPG 2008 SurfaceResults

(Measured 0‐6 inches)

Monte Carlo


Concentrations

USEPA 2008 Surface Grab (0‐1 inch) and

Sediment Trap

CPG 2008 Surface Results (Measured 0‐6 inches)

Average ± Standard

Deviation (Range) Count = 4

Average ± Standard Deviation (Range) Count = 3 (PCBs,

Metals, PAHs, Pesticides) Count = 2

(Dioxins)

Average ± Standard Deviation

(Range) Count = 3

Average ± Standard Deviation (Range)

Count = 4


Count = 4 (Dioxins, PCBs, PAHs, Pesticides) Count = 6

(Metals)


Count = 3

2,3,7,8‐TCDD ng/kg 0.1 to 1.8 0.92 ± 0.98 (0.1 to 1.8)

4.2 ± 5.9 (0.030 to 8.3) 1.7 to 2.4

2.0 ± 0.35 (1.7 to 2.5)

48 ± 83 (0.062 to 144) 1.2 to 6.7

2.9 ± 2.6 (1.2 to 6.7)

0.075 ± 0.039 (0.033 to 0.11)

Total TCDD ng/kg 3 to 22.8 11 ± 8.4

(3.0 to 23) 3.1 ± 3.2

(0.89 to 5.4) 13.3 to 30.7 24 ± 7.8

(13 to 31) 31 ± 54

(0.040 to 93) 18.7 to 34 25 ± 6.8

(19 to 34) 1.6 ± 1.2

(0.37 to 2.7)

Total PCB mg/kg 0 to 0.2 0.10 ± 0.054

(0.043 to 0.17) 0.034 ± 0.023

(0.011 to 0.056) 0.1 to 1 0.40 ± 0.43 (0.12

to 1.0) 0.092 ± 0.12

(0.0032 to 0.23) 0 to 1.1 0.37 ± 0.51

(0.050 to 1.1) 0.10 ± 0.076

(0.038 to 0.19)

Mercury mg/kg 0.1 to 0.5 0.26 ± 0.15

(0.097 to 0.46) 0.90 ± 0.045 (0.051 to

0.14) 0.3 to 0.7 0.48 ± 0.19 (0.29

to 0.67) 0.18 ± 0.13

(0.085 to 0.32) 0.1 to 0.2 0.10 ± 0.038

(0.063 to 0.17) 0.12 ± 0.11

(0.033 to 0.24)

Chromium mg/kg 18.2 to 40.9 25 ± 11

(18 to 41) 11 ± 1.6

(9.3 to 12) 27.2 to 42.9 35 ± 6.5

(27 to 43) 29 ± 14

(13 to 37) 14.8 to 36.5 20 ± 8.4

(15 to 37) 21 ± 7.9

(13 to 29)

Benzo[a]pyrene ug/kg 792 to 7,089 3,300 ± 2,700 (790 to 7,100)

220 ± 150 (70 to 370) 3,810 to 4,930

4,300 ± 500 (3,800 to 4,900)

1,300 ± 1,300 (0.018 to 2,600) 2,600 to 4,799

3,700 ± 1,100 (2,600 to 4,800)

1,100 ± 420 (670 to 1,500)

Fluoranthene ug/kg 1,770 to 17,599 8,400 ± 6,900

(1,800 to 18,000) 400 ± 350 (75 to 770)

8,220 to 11,099

9,500 ± 1,400 (8,200 to 11,000)

2,800 ± 3,100 (1.9 to 6,100) 5,832 to 11,499

8,700 ± 3,200 (5,800 to 12,000)

1,500 ± 1,400 (90 to 2,800)

4,4'‐DDE ug/kg 6.6 to 54.7 26 ± 20

(6.6 to 55) 4.2 ± 3.7

(1.6 to 8.5) 35 to 60.1 46 ± 11

(35 to 60) 15 ± 14

(0.12 to 28) 12.7 to 28 19 ± 6.8

(13 to 28) 4.7 ± 1.5

(3.7 to 6.4)

gamma-Chlordane ug/kg 12.1 to 54.2

27 ± 19 (12 to 54)

3.4 ± 0.91 (2.4 to 4.1) 33.9 to 139

77 ± 45 (34 to 140)

31 ± 28 (0.21 to 55) 31.7 to 77.6

55 ± 23 (32 to 78)

12 ± 1.0 (11 to 13)

Notes: 1. The nondetect values are presented as half the detection limit. 2. Dundee Dam samples are repeated for completeness. 3. CPG 2008 2,3,7,8‐TCDD values are presented corrected using the correction factor. 4. CPG 2008 and 2009 uses lab generated sum of PCB congeners. Measured average surface sediment concentration is higher than the simulated concentration range Measured average surface sediment concentration is less than the simulated concentration range


Page 2 of 2

Table 4‐5: Comparison of 2008 Tributary and Dundee Dam Measured Surface Sediment Concentrations with Monte Carlo Simulated Model Inputs

Dundee Dam

Contaminant

Units

Monte Carlo Model Simulated Range of

Possible Concentrations

USEPA 2008 Surface Grab (0‐1 inch) and Sediment Trap

USEPA High Resolution Cores (Measured Composite 0‐1

inch or 2 cm)

USEPA High Resolution Cores (Calculated Length Weighted Average 0‐

6.3 inches or 16 cm) CPG 2008 Surface Results

(Measured 0‐6 inches)


Count = 8 (Dioxins, PCBs, PAHs, Pesticides) Count = 10 (Metals)


Count = 4

Average ± Standard Deviation (Range) Count = 3 (Dioxins, PCBs, and PAHs)

Count = 4 (Metals and Pesticides)


Count = 6

2,3,7,8‐TCDD ng/kg 1 to 2.9 1.8 ± 0.56 (1.0 to 2.9)

2.5 ± 0.59 (1.9 to 3.0)

10 ± 11 (3.2 to 24)

6.2 ± 15 (0.038 to 37)

Total TCDD ng/kg 26 to 73.4 41 ± 16

(26 to 73) 70 ± 22

(37 to 87) 170 ± 99

(70 to 270) 150 ± 330

(4.0 to 810)

Total PCB mg/kg 0.2 to 0.7 0.34 ± 0.19

(0.21 to 0.69)0.60 ± 0.090 (0.50 to 0.69)

2.4 ± 1.8 (0.46 to 4.0)

1.1 ± 2.0 (0.083 to 5.1)

Mercury mg/kg 0.4 to 1.8 0.75 ± 0.40 (0.46 to 1.8)

1.5 ± 0.71 (0.72 to 2.2)

3.1 ± 1.4 (2.2 to 5.1)

2.4 ± 3.9 (0.40 to 10)

Chromium mg/kg 20.6 to 51.4 32 ± 9.8

(21 to 51) 41 ± 16

(23 to 58) 73 ± 17

(57 to 94) 31 ± 18

(12 to 55)

Benzo[a]pyrene ug/kg 3,471 to 9,749 5,400 ± 1,400

(3,500 to 7,200) 7,800 ± 3,800

(3,500 to 13,000) 15,000 ± 11,000 (5,700 to 29,000)

26,000 ± 27,000 (620 to 53,000)

Fluoranthene ug/kg 5,950 to 16,197 8,800 ± 1,900

(6,000 to 12,000) 7,600 ± 2,100

(6,100 to 11,000) 14,000 ± 6,000

(8,900 to 22,000) 42,000 ± 47,000 (850 to 100,000)

4,4'‐DDE ug/kg 7.5 to 25.6 13 ± 5.6

(7.5 to 26) 18 ± 5.2

(14 to 26) 42 ± 30

(6.5 to 62) 40 ± 84

(1.1 to 210)

gamma-Chlordane ug/kg 13.5 to 47 24 ± 9.9

(15 to 47) 34 ± 11

(21 to 47) 34 ± 34

(10 to 73) 11 ± 10

(1.6 to 29)

Notes: 1. The nondetect values are presented as half the detection limit. 2. Dundee Dam samples are repeated for completeness. 3. CPG 2008 2,3,7,8‐TCDD values are presented corrected using the correction factor. 4. CPG 2008 and 2009 uses lab generated sum of PCB congeners. Measured average surface sediment concentration is higher than the simulated concentration range Measured average surface sediment concentration is less than the simulated concentration range

Appendix C: Mass Balance Modeling Analysis 2014 Lower Eight Miles of the Lower Passaic River

Table 5-1: Comparison of the Average 1980s Concentrations and 2005 Surface Sediment Concentrations for Selected Contaminants Analyte Average 1980s Decadal

Concentration Average 2005 Surface Sediment

Concentration Mercury (mg/kg) 3.3 1.8 Lead (mg/kg) 320 210 Copper (mg/kg) 180 150 gamma-Chlordane (μg/kg) 85 70 Dieldrin (μg/kg) 2.4 5.8 4,4’-DDE (μg/kg) 110 54 2,3,7,8-TCDD (ng/kg)a 560 430 Total PCB (μg/kg)a 2,500 1,000 LMW PAH (mg/kg) 10 10 HMW PAH (mg/kg) 25 28 Concentrations rounded to two significant figures. Note: a: Average decadal concentration for three river locations (RM1.4, RM2.2, and RM11)

Table 5-2: Contaminant Half Times for “Excess Concentration” in Lower Passaic River Sediments Based on High Resolution Cores from 2005 and Surface Samples in 2007 (See Text for Explanation) Analyte Half Time (Confidence Interval)

1980-2007 (years) Mercury (mg/kg) 34 (19-173) Lead (mg/kg) 39.5 (23-151) Copper (mg/kg) 57 (30.7-405) gamma-Chlordane (μg/kg)a 99 4,4’-DDE (μg/kg) 19 (13-33) Dieldrin (μg/kg) No Applicableb 2,3,7,8-TCDD (ng/kg) 25 (14-101) Total PCB (μg/kg) 26 (16-61) HMW PAH (mg/kg)1 44 LMW PAH (mg/kg) 63 Note: a: No statistically significant trend with time. Future concentrations are taken as constant. b: Increasing trend with time. Half time for decline cannot be determined.

Appendix C: Mass Balance Modeling Analysis 2014 Lower Eight Miles of the Lower Passaic River

Table 5-3a: Reduction in Best Estimate Forecasted Surface Sediment Concentrations in 2059 relative to 2017 Contaminant Alternative 1 Alternative 2 Alternative 3 Alternative 4 gamma‐Chlordane 0% 38% 32% 12% Copper 23% 63% 59% 37% 4,4’‐DDE 50% 78% 75% 60% 2,3,7,8‐TCDD 59% 91% 90% 71% Mercury 37% 72% 68% 49% Lead 28% 65% 61% 41% Total PCB 48% 79% 76% 59% HMW PAH 0% 34% 27% 10%

Table 5-3b: Best Estimate Forecasted Surface Sediment Concentrations in 2017 and 2059

Contaminant

Concentrations in 2017 for

All Alternatives

Concentrations in 2059

Alternative 1

Alternative 2

Alternative 3

Alternative 4

gamma‐Chlordane 23 24 15 16 21 Copper 160 130 60 70 100 4,4’‐DDE 50 30 10 10 20 2,3,7,8‐TCDD 0.5 0.2 0.04 0.05 0.1 Mercury 2.5 1.5 0.7 0.8 1.2 Lead 260 180 90 100 150 TotalPCB 1,590 820 340 380 650 HMWPAH 43,000 43,000 28,000 31,000 38,000

Note: Concentration is rounded to 2 significant digits.


FIGURES

Lower Eight Miles of the Lower Passaic River2014

Figure 3-1Sources of Model for the Lower Passaic River Empirical Mass Balance

Legend

Notes

Lower Passaic River Newark Bay

Dundee Lake

SaddleRiver

SecondRiver

ThirdRiver CSO/SWO

Water

Sediment

Dundee Dam


Figure 3-2Relationship between Average Contaminant Concentration and Standard Error for the Recently-Deposited Sediments in the Main Stem

Lower Passaic River

R² = 0.98

1.E‐05

1.E‐04

1.E‐03

1.E‐02

1.E‐01

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

1.E‐05 1.E‐04 1.E‐03 1.E‐02 1.E‐01 1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07 1.E+08

Standard Error (ppb

)

Average Concentration (ppb)

PCDD/F

PCB

OCP

PAH

Metals


Figure 3-3Cluster Analysis for PAHs

Method = Ward

Acenaphthene

Acenaphthylene

AnthraceneBenz[a]anthraceneBenzo[a]pyreneBenzo[g,h,i]perylene

Benzo[j,k]fluoranthene

Benzo[j]fluorantheneChrysene

Dibenz[a,h]anthracene

Fluoranthene

Fluorene

Indeno[1,2,3-c,d]-pyrene

Naphthalene

Phenanthrene

Pyrene

Dendrogram

Hierarchical Clustering


Figure 3-4Cluster Analysis for Metals

Method = Ward

Cadmium

Chromium

Copper

Lead

Mercury

Nickel

Silver

Dendrogram

Hierarchical Clustering


Figure 4‐1Best Estimate of Relative Solids Contribution to the Lower Passaic River Estimated by the EMB Model

Legend

Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO

Newark Bay Northern EndResuspension (LowerPassaic River)

2014

Resuspension (Lower Passaic River)

48%

Newark Bay14%

Upper Passaic River32%

Saddle River4%

Third River1%

Second River / SWO1%CSO0%CSO0.5%


Figure 4‐2Fractional Contribution for Solids(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile

75th PercentileBest Estimate *

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution

Notes:*: Best Estimate representsMass Balance estimates using average input valuesfor source profiles.

5th, 25th, 50th, 75th and 95thpercentiles are based onMass Balance results for the 10,000 iterations in MonteCarlo Analysis.

2014

Lower Eight Miles of the Lower Passaic River 2014

Figure 4-3aSource Concentration and Mass Balance for 2,3,7,8-TCDD

Legend

Notes

Legend

*: Second River Results are used to represent the SWOs. (see Appendix F)

Source Concentration of 2,3,7,8-TCDDMaximum

Minimum

Median25% quartile

75% quartile

Outliers

Mean ± 2 standarderrors

Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO


Best Estimate Mass Balance for 2,3,7,8-TCDD

Average

0.0001

0.001

0.01

0.1

1

10

100

Upp

er P

assa

ic R

iver

Sad

dle

Riv

er

Third

Riv

er

Sec

ond

Riv

er/S

WO

*

CS

O

New

ark

Bay

N

orth

ern

End

1995

0-6

inch

S

urfa

ce S

edim

ent

2,3,

7,8-

TCD

D C

once

ntra

tion

(ug/

kg)

Lower Passaic RiverTarget Concentration(RM2 to 12)


97%

Newark Bay3%

Upper Passaic River0.1%

Saddle River0.03%

Third River0.004%

Second River / SWO0.003%

CSO0.005%


Figure 4‐3bFractional Contribution for 2,3,7,8‐TCDD(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile



5th, 25th, 50th, 75th and 95thpercentiles are based onMass Balance results for the 10,000 iterations in MonteCarlo Analysis.0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution

2014


Figure 4-4aSource Concentration and Mass Balance for Total Tetra-dioxins

Legend

Notes

Legend


Source Concentration of Total Tetra-dioxinsMaximum

Minimum

Median25% quartile

75% quartile

Outliers


Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO


Best Estimate Mass Balance for Total Tetra-dioxins

Average

0.001

0.01

0.1

1

10

100

Upp

er P

assa

ic R

iver

Sad

dle

Riv

er

Third

Riv

er

Sec

ond

Riv

er/S

WO

*

CS

O

New

ark

Bay

Nor

ther

n E

nd

1995

0-6

inch

S

urfa

ce S

edim

ent

Tota

l Tet

ra-d

ioxi

ns C

once

ntra

tion

(ug/

kg)

Lower Passaic RiverTarget Concentration(RM2 to 12)


92%

Newark Bay5%


Saddle River0.2%

Third River0.04%


CSO0.07%


Figure 4‐4bFractional Contribution for Total Tetra‐Dioxins(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile




0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution

2014


Figure 4-5aSource Concentration and Mass Balance for Total PCBs

Legend

Notes

Legend


Source Concentration of Total PCBsMaximum

Minimum

Median25% quartile

75% quartile

Outliers


Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO


Best Estimate Mass Balance for Total PCBs

Average

10

100

1,000

10,000

100,000

Upp

er P

assa

ic R

iver

Sad

dle

Riv

er

Third

Riv

er

Seco

nd R

iver

/SW

O*

CS

O

New

ark

Bay

Nor

ther

n En

d

1995

0-6

inch

Su

rface

Sed

imen

t

Tota

l PC

Bs

(ug/

kg)

Target Concentration Lower Passaic Average Surface Sediment (RM2 to 12)


81%

Newark Bay7%


Saddle River1%

Third River0.3%


CSO0.4%


Figure 4‐5bFractional Contribution for Total PCBs(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile




0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution

2014


Figure 4-6aSource Concentration and Mass Balance for Benzo[a]pyrene

Legend

Notes

Legend


Source Concentration of Benzo[a]pyreneMaximum

Minimum

Median25% quartile

75% quartile

Outliers


Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO


Best Estimate Mass Balance for Benzo[a]pyrene

Average

100

1,000

10,000

100,000

Upp

er P

assa

ic R

iver

Sad

dle

Riv

er

Third

Riv

er

Sec

ond

Riv

er/S

WO

*

CSO

New

ark

Bay

Nor

ther

n En

d

1995

0-6

inch

S

urfa

ce S

edim

ent

Benz

o[a]

pyre

ne C

once

ntra

tion

(ug/

kg)

Lower Passaic RiverTarget ConcentrationRM2 to 12


33%

Newark Bay7%


Saddle River4%

Third River1%

Second River / SWO2%

CSO0.3%


Figure 4‐6bFractional Contribution for Benzo[a]pyrene(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile

75th PercentileBest Estimate*



0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution

2014


Figure 4-7aSource Concentration and Mass Balance for Fluoranthene

Legend

Notes

Legend


Source Concentration of FluorantheneMaximum

Minimum

Median25% quartile

75% quartile

Outliers


Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO


Best Estimate Mass Balance for Fluoranthene

Average

100

1,000

10,000

100,000

Upp

er P

assa

ic R

iver

Sad

dle

Riv

er

Third

Riv

er

Seco

nd R

iver

/SW

O*

CSO

New

ark

Bay

Nor

ther

n E

nd

1995

0-6

inch

Su

rface

Sed

imen

t

Fluo

rant

hene

Con

cent

ratio

n (u

g/kg

)

Lower Pasaic RiverTarget ConcentrationRM2 to 12


40%

Newark Bay5%


Saddle River5%

Third River1%


CSO0.4%


Figure 4‐7bFractional Contribution for Fluoranthene(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile




0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution

2014


Figure 4-8aSource Concentration and Mass Balance for 4,4’-DDE

Legend

Notes

Legend


Source Concentration of 4,4’-DDEMaximum

Minimum

Median25% quartile

75% quartile

Outliers


Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO


Best Estimate Mass Balance for 4,4’-DDE

Average

1

10

100

1,000

Upp

er P

assa

ic R

iver

Sad

dle

Riv

er

Third

Riv

er

Sec

ond

Riv

er/S

WO

*

CS

O

New

ark

Bay

N

orth

ern

End

1995

0-6

inch

S

urfa

ce S

edim

ent

4,4'

-DD

E C

once

ntra

tion

(ug/

kg) Lower Passaic River

Target ConcentrationRM2 to 12


78%

Newark Bay8%


Saddle River2%

Third River1%


CSO0.3%


Figure 4‐8bFractional Contribution for 4,4’‐DDE(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile




0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution

2014


Figure 4-9aSource Concentration and Mass Balance for gamma-Chlordane

Legend

Notes

Legend

*Second River Results are used to represent the SWOs. (see Appendix F)+Robinson (2002)

Source Concentration of gamma-ChlordaneMaximum

Minimum

Median25% quartile

75% quartile

Outliers


Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO


Best Estimate Mass Balance for gamma-Chlordane

Average


52%

Newark Bay3%


Saddle River8%

Third River2%


CSO1%

1

10

100

1,000

Upp

er P

assa

ic R

iver

Sadd

le R

iver

Third

Riv

er

Seco

nd R

iver

/SW

O*

CS

O

New

ark

Bay

Nor

ther

n En

d+

1995

0-6

inch

Su

rface

Sed

imen

t

Chl

orda

ne (g

amm

a) (u

g/kg

)



Figure 4‐9bFractional Contribution for gamma‐Chlordane(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile




0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution

2014


Figure 4-10aSource Concentration and Mass Balance for Copper

Legend

Notes

Legend


Source Concentration of CopperMaximum

Minimum

Median25% quartile

75% quartile

Outliers


Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO


Best Estimate Mass Balance for Copper

Average

10

100

1,000

Upp

er P

assa

ic R

iver

Sad

dle

Riv

er

Third

Riv

er

Sec

ond

Riv

er/S

WO

*

CS

O

New

ark

Bay

N

orth

ern

End

1995

0-6

inch

S

urfa

ce S

edim

ent

Cop

per C

once

ntra

tion

(mg/

kg)


1 Outlier (2,470 mg/kg)


72%

Newark Bay12%


Saddle River1%

Third River0.3%


CSO1%


Figure 4‐10bFractional Contribution for Copper(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile




0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution

2014


Figure 4-11aSource Concentration and Mass Balance for Chromium

Legend

Notes

Legend


Source Concentration of ChromiumMaximum

Minimum

Median25% quartile

75% quartile

Outliers


Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO


Best Estimate Mass Balance for Chromium

Average

10

100

1,000

Upp

er P

assa

ic R

iver

Sad

dle

Riv

er

Third

Riv

er

Sec

ond

Riv

er/S

WO

*

CS

O

New

ark

Bay

N

orth

ern

End

1995

0-6

inch

Su

rface

Sed

imen

t

Chr

omiu

m C

once

ntra

tion

(mg/

kg)


1 Outlier (7.9 mg/kg)


74%

Newark Bay15%


Saddle River1%

Third River0.3%


CSO0.3%


Figure 4‐11bFractional Contribution for Chromium(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile




0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution

2014


Figure 4-12aSource Concentration and Mass Balance for Mercury

Legend

Notes

Legend


Source Concentration of MercuryMaximum

Minimum

Median25% quartile

75% quartile

Outliers


Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO


Best Estimate Mass Balance for Mercury

Average

0.01

0.1

1

10

100

Upp

er P

assa

ic R

iver

Sad

dle

Riv

er

Third

Riv

er

Sec

ond

Riv

er/S

WO

*

CS

O

New

ark

Bay

N

orth

ern

End

1995

0-6

inch

S

urfa

ce S

edim

ent

Mer

cury

Con

cent

ratio

n (m

g/kg

) Lower Passaic RiverTarget ConcentrationRM2 to 12


75%

Newark Bay14%


Saddle River0.2%

Third River0.2% Second River / SWO

0.2%CSO0.2%


Figure 4‐12bFractional Contribution for Mercury(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile




0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution

2014


Figure 4-13aSource Concentration and Mass Balance for Lead

Legend

Notes

Legend


Source Concentration of LeadMaximum

Minimum

Median25% quartile

75% quartile

Outliers


Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO


Best Estimate Mass Balance for Lead

Average

10

100

1,000

Upp

er P

assa

ic R

iver

Sad

dle

Riv

er

Third

Riv

er

Seco

nd R

iver

/SW

O*

CSO

New

ark

Bay

N

orth

ern

End

1995

0-6

inch

Su

rface

Sed

imen

t

Lead

Con

cent

ratio

n (m

g/kg

)


2 Outliers (4.4 mg/kg, 5.6 mg/kg)


71%

Newark Bay7%


Saddle River1%

Third River0.5%

Second River / SWO1%CSO1%


Figure 4‐13bFractional Contribution for Lead(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile




0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution

2014


Figure 4-14aSource Concentration and Mass Balance for Iron

Legend

Notes

Legend


Source Concentration of IronMaximum

Minimum

Median25% quartile

75% quartile

Outliers


Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO


Best Estimate Mass Balance for Iron

1,000

10,000

100,000

Upp

er P

assa

ic R

iver

Sad

dle

Riv

er

Third

Riv

er

Sec

ond

Riv

er/S

WO

*

CS

O

New

ark

Bay

N

orth

ern

End

1995

0-6

inch

S

urfa

ce S

edim

ent

Iron

Con

cent

ratio

n (m

g/kg

)


Average


54%

Newark Bay18%


Saddle River2%

Third River1%


CSO0.5%


Figure 4‐14bFractional Contribution for Iron(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile




0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution

2014


Figure 4-15aSource Concentration and Mass Balance for Total Organic Carbon

Legend

Notes

Legend


Source Concentration of Total Organic CarbonMaximum

Minimum

Median25% quartile

75% quartile

Outliers


Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO


Best Estimate Mass Balance for Total Organic Carbon

100

1,000

10,000

100,000

1,000,000

Upp

er P

assa

ic R

iver

Sad

dle

Riv

er

Third

Riv

er

Seco

nd R

iver

/SW

O*

CS

O

New

ark

Bay

N

orth

ern

End

1995

0-6

inch

S

urfa

ce S

edim

ent

Tota

l Org

anic

Car

bon

(mg/

kg)


Average


72%

Newark Bay5%


Saddle River2%

Third River1%


CSO2%


Figure 4‐15b

2014

Fractional Contribution for Total Organic Carbon(Results from Monte Carlo Analysis and Best Estimate Solution)

Legend

95th Percentile

5th Percentile

50th Percentile

25th Percentile




0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Resuspen

sion

New

ark B

ay

Upp

er Passaic River

Sadd

le River

Second

River/SWO

Third

River

CSO

Fractio

nal C

ontribution


Figure 4‐16

2014

Dieldrin Contribution to the Lower Passaic River

Legend

Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO



73%

Newark Bay7%


Saddle River5%

Third River1%


CSO1%


Figure 4‐17

2014

Phenanthrene Contribution to the Lower Passaic River

Legend

Upper Passaic River

Saddle River

Second River/SWO

Third River

CSO



47%

Newark Bay2%


Saddle River4%

Third River1%


CSO2%


Figure 4‐18Evaluation of EMB Model Performance

-0.50

-0.40

-0.30

-0.20

-0.10

0.00

0.10

0.20

0.30

0.40

0.50

Ars

enic

Cob

alt

Zinc

Nic

kel

Cop

per

Chr

omiu

m

Iron

Mer

cury

Lead

Cad

miu

m

Chl

orda

ne

DD

E

2,3,

7,8-

TCD

D

Tota

l TC

DD

Tota

l PC

B

Ben

z(a)

anth

race

ne

Ben

zo(a

)pyr

ene

Chr

ysen

e

Fluo

rant

hene

Inde

no(1

,2,3

-cd)

pyre

ne

Pyr

ene

TOC

Norm

aliz

ed M

ean

Erro

r (N

ME)

Legend‐‐‐‐‐‐‐ 25%Chemical directly included in model.

Chemical used to evaluate optimized model.

2014

gam

ma-

Chl

orda

ne


0.1

1

1980 1985 1990 1995 2000 2005 2010

Deposition Year

Excess 2,3,7,8‐TCD

D Con

centratio

n (ug/kg)

Figure 5-1Excess 2,3,7,8-TCDD Concentration vs. Approximate Year of Deposition

Note:Excess concentration represents the difference between the measured concentration in the Lower Passaic River recently deposited (Be-7 bearing) samples and the baseline concentration. The excess is attributed to resuspension and Newark Bay derived loads.

R2 = 0.168 (p = 0.01)

= -0.0271

Half-Time = 25 years (Confidence Interval: 14 years – 101 years)

RM1.4 RM2.2RM7.8RM11RM12.6Exponential Trend Line

Approximate Year of Deposition


Figure 5-2Excess Mercury Concentration vs. Approximate Year of Deposition



0.1

1

10

1980 1985 1990 1995 2000 2005 2010

Exce

ss M

ercu

ry C

once

ntra

tion

(ug/

kg)

R2 = 0.0975 = -0.0201Half -Life = 34 years



R2 = 0.0975 (p = 0.015)

= -0.02



Figure 5-3Excess Lead Concentration vs. Approximate Year of Deposition


10

100

1000

1980 1985 1990 1995 2000 2005 2010

Exce

ss L

ead

Con

cent

ratio

n (u

g/kg

)

R2 = 0.116 = -0.0175Half -Life = 40 years



R2 = 0.116 (p <0.01)

= -0.0175

Half-Time = 39.5 years (Confidence Interval: 23 years – 151 years)


Figure 5-4Excess Copper Concentration vs. Approximate Year of Deposition


10

100

1000

1980 1985 1990 1995 2000 2005 2010

Exce

ss C

oppe

r Con

cent

ratio

n (u

g/kg

)

R2 = 0.0837 = -0.0122Half -Life = 57 years



R2 = 0.084 (p = 0.02)

= -0.01

Half-Time = 57 years (Confidence Interval: 30.7 years – 405 years)


10

100

1,000

1980 1985 1990 1995 2000 2005 2010

Deposition Year

Excess 4,4'‐D

DE Co

ncen

tration (ug/kg)

Figure 5-5Excess 4,4’-DDE Concentration vs. Approximate Year of Deposition


R2 = 0.407 (p < 0.01)

= -0.036





Figure 5-6Excess gamma-Chlordane Concentration vs. Approximate Year of Deposition


1

10

100

1980 1985 1990 1995 2000 2005 2010

Exce

ss g

amm

a-C

hlor

dane

(tra

ns) C

once

ntra

tion

(ug/

kg)

R2 = -0.0116 = -0.00701Half -Life = 99 years



R2 = -0.01 (p = 0.44)

= -0.007

Half-Time = 99 years


0.01

0.1

1

10

100

1980 1985 1990 1995 2000 2005 2010

Deposition Year

Excess Dieldrin

Con

centratio

n (ug/kg)

Figure 5-7Excess Dieldrin Concentration vs. Approximate Year of Deposition


Increasing Trend

No Half-Time




100

1,000

10,000

1980 1985 1990 1995 2000 2005 2010

Deposition Year

Excess Total PCB

Con

centratio

n (ug/kg)

Figure 5-8Excess Total PCBs Concentration vs. Approximate Year of Deposition


R2 = 0.25 (p < 0.01)

= -0.027





1,000

10,000

100,000

1980 1985 1990 1995 2000 2005 2010

Deposition Year

Excess HMW PAH

Con

centratio

n (ug/kg)

Figure 5-9Excess High Molecular Weight PAH Concentration vs. Approximate Year of Deposition


R2 = 0.06 (p = 0.06)





Figure 5-10Excess Low Molecular Weight PAH Concentration vs. Approximate Year of Deposition


R2 = 0.19



1,000

10,000

1980 1985 1990 1995 2000 2005 2010

LMW PAH M

easure Con

centration

(ug/kg)

Mid Year

R2 = 0.082 Half‐Life = 63 years

Note that the concentration is not the excess concentration, but the measured concentration for each core



R2 = 0.07 (p = 0.04)


2014Lower Eight Miles of the Lower Passaic River

Figure 5-11a2,3,7,8-TCDD Best Estimates of Trajectories for 0-6 inch Biologically Active Layer

Legend

NoteBest Estimate represents trajectory results using average input values for all parameters.

Alternative 4

Alternative 3

Alternative 2

Alternative 1

1995 TSI Data Set

1995 TSI Dataset

Mean

+ 2 Standard Errors

- 2 Standard Errors

0.001

0.01

0.1

1

10

1990 2000 2010 2020 2030 2040 2050 2060

2,3,

7,8-

TC

DD

Con

cent

ratio

n (u

g/kg

)

Year


0.001

0.01

0.1

1

10

0.001

0.01

0.1

1

10

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

2,3,

7,8-

TC

DD

Con

cent

ratio

n in

Sur

face

Sed

imen

t (0-

6 in

ch)

(ug/

kg)

Year

Figure 5-11bMonte Carlo Trajectory Results for 2,3,7,8-TCDD Concentration in 0-6 inch Surface Sediment

(Alternative 1 and Alternative 2)

Alternative 1 and Alternative 2 LegendAlternative 1

Alternative 2


25th to 75th Percentile5th to 95th Percentile

Best Estimate


Best Estimate

1995 TSI DatasetCPG Dataset

Mean

+ 2 Standard Errors

- 2 Standard Errors


Figure 5-11cMonte Carlo Trajectory Results for 2,3,7,8-TCDD Concentration in 0-6 inch Surface Sediment



Alternative 3



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0.001

0.01

0.1

1

10

0.001

0.01

0.1

1

10

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

2,3,

7,8-

TC

DD

Con

cent

ratio

n in

Sur

face

Sed

imen

t (0-

6 in

ch)

(ug/

kg)

Year


Figure 5-11dMonte Carlo Trajectory Results for 2,3,7,8-TCDD Concentration in 0-6 inch Surface Sediment



Alternative 4



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0.001

0.01

0.1

1

10

0.001

0.01

0.1

1

10

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

2,3,

7,8-

TC

DD

Con

cent

ratio

n in

Sur

face

Sed

imen

t (0-

6 in

ch)

(ug/

kg)

Year


Figure 5-12aMercury Best Estimates of Trajectories for 0-6 inch Biologically Active Layer

Legend


1995 TSI Dataset

Mean

+ 2 Standard Errors

- 2 Standard Errors

Alternative 4

Alternative 3

Alternative 2

Alternative 1

1995 TSI Data Set

0.1

1

10

1990 2000 2010 2020 2030 2040 2050 2060

Mer

cury

Con

cent

ratio

n (m

g/kg

)

Year


Figure 5-12bMonte Carlo Trajectory Results for Mercury Concentration in 0-6 inch Surface Sediment



Alternative 2



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

Mer

cury

Con

cent

ratio

n in

Sur

face

Sed

imen

t (0-

6 in

ch)

(mg/

kg)

Year


Figure 5-12cMonte Carlo Trajectory Results for Mercury Concentration in 0-6 inch Surface Sediment



Alternative 3



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

Mer

cury

Con

cent

ratio

n in

Sur

face

Sed

imen

t (0-

6 in

ch)

(mg/

kg)

Year


Figure 5-12dMonte Carlo Trajectory Results for Mercury Concentration in 0-6 inch Surface Sediment



Alternative 4



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

Mer

cury

Con

cent

ratio

n in

Sur

face

Sed

imen

t (0-

6 in

ch)

(mg/

kg)

Year


Figure 5-13aLead Best Estimates of Trajectories for 0-6 inch Biologically Active Layer

Legend


1995 TSI Dataset

Mean

+ 2 Standard Errors

- 2 Standard Errors

Alternative 4

Alternative 3

Alternative 2

Alternative 1

1995 TSI Data Set

10

100

1,000

1990 2000 2010 2020 2030 2040 2050 2060

Lea

d C

once

ntra

tion

(mg/

kg)

Year


Figure 5-13bMonte Carlo Trajectory Results for Lead Concentration in 0-6 inch Surface Sediment



Alternative 2



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

50

100

150

200

250

300

350

400

0

50

100

150

200

250

300

350

400

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

Lea

d C

once

ntra

tion

in S

urfa

ce S

edim

ent (

0-6

inch

)(m

g/kg

)

Year


Figure 5-13cMonte Carlo Trajectory Results for Lead Concentration in 0-6 inch Surface Sediment



Alternative 3



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

50

100

150

200

250

300

350

400

0

50

100

150

200

250

300

350

400

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

Lea

d C

once

ntra

tion

in S

urfa

ce S

edim

ent (

0-6

inch

)(m

g/kg

)

Year


Figure 5-13dMonte Carlo Trajectory Results for Lead Concentration in 0-6 inch Surface Sediment



Alternative 4



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

50

100

150

200

250

300

350

400

0

50

100

150

200

250

300

350

400

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

Lea

d C

once

ntra

tion

in S

urfa

ce S

edim

ent (

0-6

inch

)(m

g/kg

)

Year


Figure 5-14aCopper Best Estimates of Trajectories for 0-6 inch Biologically Active Layer

Legend


1995 TSI Dataset

Mean

+ 2 Standard Errors

- 2 Standard Errors

Alternative 4

Alternative 3

Alternative 2

Alternative 1

1995 TSI Data Set

1

10

100

1,000

1990 2000 2010 2020 2030 2040 2050 2060

Cop

per

Con

cent

ratio

n (m

g/kg

)

Year


Figure 5-14bMonte Carlo Trajectory Results for Copper Concentration in 0-6 inch Surface Sediment



Alternative 2



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

50

100

150

200

250

300

0

50

100

150

200

250

300

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

Cop

per

Con

cent

ratio

n in

Sur

face

Sed

imen

t (0-

6 in

ch)

(mg/

kg)

Year


Figure 5-14cMonte Carlo Trajectory Results for Copper Concentration in 0-6 inch Surface Sediment



Alternative 3



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

50

100

150

200

250

300

0

50

100

150

200

250

300

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

Cop

per

Con

cent

ratio

n in

Sur

face

Sed

imen

t (0-

6 in

ch)

(mg/

kg)

Year


Figure 5-14dMonte Carlo Trajectory Results for Copper Concentration in 0-6 inch Surface Sediment



Alternative 4



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

50

100

150

200

250

300

0

50

100

150

200

250

300

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

Cop

per

Con

cent

ratio

n in

Sur

face

Sed

imen

t (0-

6 in

ch)

(mg/

kg)

Year


Figure 5-15a4,4’-DDE Best Estimates of Trajectories for 0-6 inch Biologically Active Layer

Legend


1995 TSI Dataset

Mean

+ 2 Standard Errors

- 2 Standard Errors

Alternative 4

Alternative 3

Alternative 2

Alternative 1

1995 TSI Data Set

1

10

100

1990 2000 2010 2020 2030 2040 2050 2060

4,4'

-DD

E C

once

ntra

tion

(ug/

kg)

Year


Figure 5-15bMonte Carlo Trajectory Results for 4,4’-DDE Concentration in 0-6 inch Surface Sediment



Alternative 2



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

20

40

60

80

100

120

0

20

40

60

80

100

120

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

4,4'

-DD

E C

once

ntra

tion

in S

urfa

ce S

edim

ent (

0-6

inch

)(u

g/kg

)

Year


Figure 5-15cMonte Carlo Trajectory Results for 4,4’-DDE Concentration in 0-6 inch Surface Sediment



Alternative 3



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

20

40

60

80

100

120

0

20

40

60

80

100

120

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

4, 4

'-DD

E C

once

ntra

tion

in S

urfa

ce S

edim

ent (

0-6

inch

)(u

g/kg

)

Year


Figure 5-15dMonte Carlo Trajectory Results for 4,4’-DDE Concentration in 0-6 inch Surface Sediment



Alternative 4



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

20

40

60

80

100

120

0

20

40

60

80

100

120

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

4,4'

-DD

E C

once

ntra

tion

in S

urfa

ce S

edim

ent (

0-6

inch

)(u

g/kg

)

Year


Figure 5-16agamma-Chlordane Best Estimates of Trajectories for 0-6 inch Biologically Active Layer

Legend


1995 TSI Dataset

Mean

+ 2 Standard Errors

- 2 Standard Errors

Alternative 4

Alternative 3

Alternative 2

Alternative 1

1995 TSI Data Set

1

10

100

1990 2000 2010 2020 2030 2040 2050 2060

Gam

ma-

Chl

orda

ne C

once

ntra

tion

(ug/

kg)

Year


Figure 5-16bMonte Carlo Trajectory Results for gamma-Chlordane Concentration in 0-6 inch Surface Sediment



Alternative 2



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

5

10

15

20

25

30

35

40

0

5

10

15

20

25

30

35

40

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060Gam

ma-

Chl

orda

ne C

once

ntra

tion

in S

urfa

ce S

edim

ent (

0-6

inch

)(u

g/kg

)

Year


Figure 5-16cMonte Carlo Trajectory Results for gamma-Chlordane Concentration in 0-6 inch Surface Sediment



Alternative 3



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

5

10

15

20

25

30

35

40

0

5

10

15

20

25

30

35

40

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060Gam

ma-

Chl

orda

ne C

once

ntra

tion

in S

urfa

ce S

edim

ent (

0-6

inch

)(u

g/kg

)

Year


Figure 5-16dMonte Carlo Trajectory Results for gamma-Chlordane Concentration in 0-6 inch Surface Sediment



Alternative 4



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

5

10

15

20

25

30

35

40

0

5

10

15

20

25

30

35

40

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060Gam

ma-

Chl

orda

ne C

once

ntra

tion

in S

urfa

ce S

edim

ent (

0-6

inch

)(u

g/kg

)

Year


Figure 5-17aTotal PCB Best Estimates of Trajectories for 0-6 inch Biologically Active Layer

Legend


1995 TSI Dataset

Mean

+ 2 Standard Errors

- 2 Standard Errors

Alternative 4

Alternative 3

Alternative 2

Alternative 1

1995 TSI Data Set

1995 TSI Dataset (converted from Aroclors to Congeners)

10

100

1000

10000

1990 2000 2010 2020 2030 2040 2050 2060

Tota

l PC

B C

once

ntra

tion

(ug/

kg)

Year


Figure 5-17bMonte Carlo Trajectory Results for Total PCB Concentration in 0-6 inch Surface Sediment



Alternative 2



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

500

1000

1500

2000

2500

3000

0

500

1000

1500

2000

2500

3000

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

Tota

l PC

B C

once

ntra

tion

in S

urfa

ce S

edim

ent (

0-6

inch

)(u

g/kg

)

Year


Figure 5-17cMonte Carlo Trajectory Results for Total PCB Concentration in 0-6 inch Surface Sediment



Alternative 3



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

500

1000

1500

2000

2500

3000

0

500

1000

1500

2000

2500

3000

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

Tota

l PC

B C

once

ntra

tion

in S

urfa

ce S

edim

ent (

0-6

inch

)(u

g/kg

)

Year


Figure 5-17dMonte Carlo Trajectory Results for Total PCB Concentration in 0-6 inch Surface Sediment



Alternative 4



Best Estimate


Best Estimate


Mean

+ 2 Standard Errors

- 2 Standard Errors

0

500

1000

1500

2000

2500

3000

0

500

1000

1500

2000

2500

3000

1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060

Tota

l PC

B C

once

ntra

tion

in S

urfa

ce S

edim

ent (

0-6

inch

)(u

g/kg

)

Year


Figure 5-18HMW PAH Best Estimates of Trajectories for 0-6 inch Biologically Active Layer

Legend


Alternative 4

Alternative 3

Alternative 2

Alternative 1

1995 TSI Data Set

1995 TSI Dataset

Mean

+ 2 Standard Errors

- 2 Standard Errors

1,000

10,000

100,000

1990 2000 2010 2020 2030 2040 2050 2060

HM

W P

AH

Con

cent

ratio

n (u

g/kg

)

Year


ATTACHMENTS

ATTACHMENT A

MONTE CARLO METHODOLOGY FOR UNCERTAINTY ANALYSIS

ON THE EMB MODEL AND FORECAST TRAJECTORIES

Attachment A: Monte Carlo Methodology for Uncertainty Analysis on EMB Model and Forecasts Trajectory.

1.0 Introduction

Environmental systems generally have several sources of uncertainties, and these uncertainties are not only due to a lack of proper measurements, but also due to the randomness inherent in real ecosystems. Incorporating these uncertainties into the modeling process could potentially result in providing useful information that can aid in decision-making.

The EMB model best estimate scenario assumed average values for all model inputs in determining the solids contribution, fate and transport of chemicals, as well as the forecast of future surface sediment concentrations under various remedial scenarios. To incorporate uncertainties in model parameters, a Monte Carlo1 sampling approach was used to develop 10,000 iterations of each input. These 10,000 inputs were optimized in the EMB model and the optimized results were carried through the trajectory forecast calculations. A combination Microsoft Excel® Solver and the Crystal Ball® 7 (Decisioneering, Denver, CO, USA) add-on for Microsoft Excel® (a tool typically used for solving optimization problems), was used to perform this analysis. The objective of the uncertainty analysis was to provide an insight into the level of confidence in the model estimates for the best estimate scenario. This attachment presents the detailed methodology for the Monte Carlo analysis for the EMB model and Trajectory forecasts.

2.0 Methodology The following stages were involved in the uncertainty analysis of the solids and contaminant mass balances, and contaminant forecasts presented in the Appendix C: (a) characterization of uncertainties in EMB model input chemical profiles, (b) estimation of the uncertainty in EMB model optimized outputs resulting from the uncertainty in chemical profiles, and (c) characterization of the uncertainties in model forecast resulting from uncertainties in the input profiles, EMB model outputs of solids contribution, decay of excess contaminant concentrations (lambda), and depth of resuspension reservoir/mixed layer (uncertainty propagation). A schematic diagram illustrating the Monte Carlo methodology is given in Figure A-1 and detailed description is presented below.

1 Monte Carlo simulation is categorized as a sampling method in which the trails or realizations are randomly generated from probability distributions to simulate the process of sampling from an actual population.

Attachment A: Monte Carlo Methodology for Uncertainty Analysis 2014 On the EMB Model and Forecasts Trajectory A-1 Lower Eight Miles of the Lower Passaic River

2.1 Uncertainties in EMB Model chemical input profiles Thirteen chemicals (copper, chromium, mercury, lead, gamma-Chlordane, 4,4’-DDE, 2,3,7,8 TCDD, Total TCDD, Total PCB, benzo(a)pyrene, fluoranthene, iron, and TOC) were optimized in the EMB model to determine the solids balance. The uncertainties in the concentrations of these 13 chemicals for the external sources, and the resuspension source were defined by parametric and non-parametric statistics, respectively. These are described below.

2.1.1 External Sources and Receptor Profiles The observed concentrations for the 13 chemicals for the external sources (Upper Passaic River, Newark Bay, Saddle River, Second River/SWO, Third River, and CSOs) were generally normally distributed. For each external source and the receptor, a bounded normal distribution defined by the mean, standard deviation, minimum, and maximum of each chemical was used to perform Monte Carlo simulation in Crystal Ball® 7. In performing these Monte Carlo simulations, it was important to maintain the relationship amongst the variables. Therefore, for each source, the correlation matrix was also specified in Crystal Ball® 7 to ensure that the 10,000 iterations of chemical profile represented the variability, inter-dependencies, and uncertainty for each external source and the receptor. Figure A-2a though A-2g presents the statistical distributions of chemical concentrations for the 13 chemicals optimized in the EMB model, for the external sources and the receptor.

2.1.2 Resuspension Source Profiles The chemical profiles for the resuspension source were generated based on the TSI 1995 observations. The concentrations of each chemical in this data were neither normal nor log-normal distributed. None of the complex parametric distributions in Crystal Ball® 7 could adequately fit the data set. Therefore, to create the 10,000 iterations of concentrations for the resuspension source profile, a non-parametric simulation method called a bootstrap2 was used.

The basic bootstrap approach uses Monte Carlo sampling to generate an empirical estimate of the sampling distribution of interest. In the bootstrap method, the 1995 data set was treated as the population and a Monte Carlo-style procedure was conducted on it to 10,000 iterations of the mean of the 13 chemicals optimized. This was done as follows:

2 Bootstrap is a powerful Monte Carlo method that re-samples the original sample set with replacement to generate a distribution of sample's statistics. It is a non-parametric method.


1. The original sample locations, totaling 92 from the 1995 TSI data set were assumed to define the population of data set in surface sediments for resuspension. Note that in performing this analysis, TSI Location 246 was removed from the data set because the PAH concentrations at this location were not representative of PAH values generally reported in the 1995 TSI data set.

2. The original locations were re-sampled with replacement to generate a bootstrap sample of size 91. This creates a bootstrap data set of the same size as the original, excluding Location 246. By re-sampling the locations rather than each chemical independently, the correlations amongst the chemicals were maintained. Note that this bootstrap sample set may include some sample numbers in the original sample several times, and at the same time other sample numbers may be excluded.

3. Using the chemical concentrations for the locations selected in the 91 bootstrap samples, the average concentration for each chemical was calculated.

4. Steps 2 and 3 were repeated 10,000 times to generate the empirical distribution of the resuspension source profile (Figure A-2h).

The 10,000 average concentrations generated for each chemical via bootstrap for resuspension were used along with the 10,000 iterations for the external sources and receptor to represent the uncertainty in the inputs for EMB model optimization.

2.2 Estimation of uncertainty in EMB Model Output A Microsoft Excel® macro3, which calls the SOLVER routine, was developed to perform the EMB model optimization with the aim of determining the relative solids contributions from the various sources and the mass balance for the chemicals optimized. The macro was used to solve the 10,000 optimizations using the 10,000 iterations of the sources and receptor generated by the Monte Carlo simulation. The results of the optimization run were used to understand the uncertainty in the relative source contributions and chemical mass balance for the Lower Passaic River. The 10,000 EMB model optimized results were also used as input to the trajectory forecast calculations.

3 A Microsoft Excel® macro is a set of instructions written in Visual Basic programming language for Application that can be triggered by a keyboard shortcut, toolbar button or an icon in a spreadsheet. Macros are used to eliminate the need to repeat the steps of common tasks over and over.


2.3 Uncertainties in Trajectory Forecast Uncertainties in forecasted chemical concentrations were defined by the results of 10,000 iterations of forecasted values. The chemicals forecasted included: 2,3,7,8-TCDD, mercury, copper, lead, 4,4’-DDE, Total PCB, and gamma-Chlordane. Four remedial alternatives were considered including: No Action (Alternative 1), Deep Dredging with Backfill (Alternative 2), Capping with Dredging for Flooding and Navigation (Alternative 3), and Focused Capping with Dredging for Flooding (Alternative 4). Complete descriptions of these alternatives are provided in FFS Chapter 4 and the assumptions made in the trajectories are given in Appendix C. Forecasting the future concentrations of chemicals under the various remedial scenarios required inputs of (i) chemical concentrations, (ii) solids contributions for the various sources determined by the EMB model optimization, (iii) decay of excess contaminant concentrations (lambda; λ), net sedimentation rate, and (iv) the depth of the sediment mixed layer. Uncertainties in these inputs were defined as follows:

1. Uncertainties in the chemical concentrations were defined by the 10,000 iterations used as inputs to the EMB model optimizations (Figure A-2). For each forecast calculation, the source and receptor profiles were represented by the Monte Carlo generated values as described in Section 2.1 above.

2. Uncertainties in solids contribution from the various sources were obtained from the uncertainty in the solids contributions determined by the EMB model optimization results. This was implemented by using the 10,000 solids contribution results from the EMB model.

3. Uncertainties in decay of excess sediment contamination were defined by the regression between the natural logarithm of the excess concentrations versus time (see Figure C-5-1 to C-5-10 in Appendix C). Using the slope (λ), standard error, and confidence bounds from the regressions, 10,000 iterations of λ were simulated using Monte Carlo sampling from bounded normal distributions (Figure A-3). Note: 𝐻𝑎𝑙𝑓 𝑡𝑖𝑚𝑒 = ln (0.5) 𝜆⁄

4. Uncertainties in the sedimentation rates were generated by bootstrap analysis of the differences between the 1989 and 2007 bathymetric surfaces (Figure A-4).

5. The uncertainties in depth of the sediment mixed layer were generated by 10,000 random numbers between 10 cm to 20 cm in Microsoft Excel®. Note that Microsoft Excel®’s random number generates uniform distributions of the parameter of interest (Figure A-5).

The Microsoft Excel® spreadsheets designed to perform forecast calculations using the best estimate for all inputs were modified to perform the calculations for the 10,000 iterations through


a macro. For each iteration, the macro reads the input values of chemical concentrations, λ, sediment deposition rate and mixed layer depth, updates the forecast spreadsheet with these values, and then saves the results of the forecast calculation for all the remedial scenarios.

3.0 Results Uncertainties in the EMB model solution and trajectory forecasts were defined by the confidence interval (5th and 95th percentiles) of the 10,000 optimized solutions. All the results are presented and discussed in Appendix C.


Lower Eight Miles of the Lower Passaic RiverSeptember 2008

Figure A-1Schematic Diagram of Monte Carlo Methodology for Uncertainty Analysis on EMB Model and Trajectory Forecasts

2014

10,000 iteration of chemical concentrations for sources and receptor (simulated from bounded normal distribution for external sources and receptor, bootstrap distribution for resuspension)

10,000 iteration of decay of excess sediment concentrations (l; simulated from bounded normal distribution)

10,000 iteration of sedimentation rate (simulated from bootstrap of the difference in 1989 and 2007 bathymetric surfaces)

10,000 iteration of sediment mixed layer (simulated randomly from 10 to 20 cm)

10,000 EMB Model Output. Used to define uncertainties in solids and chemical mass balance.

EMB MODEL SOLVER OPTIMIZATION

TRAJECTORY FORECAST CALCULATIONS

10,000 forecasts of chemical concentrations. Used to define uncertainties in trajectory forecasts


Figure A-2aMonte Carlo Simulation of 10,000 Iterations of Chemical Profile for Lower Passaic River (Receptor) Assuming Bounded Normal

Distributions

2600

2800

3000

3200

3400

3600

3800

4000

4200

4400

4600

4800

5000

5200

Benzo(a)pyrene

30

40

50

trans-Chlordane

90000

100000

110000

120000

130000

140000

150000

Chromium

120000

130000

140000

150000

160000

170000

180000

190000

200000

Copper

Lower Passaic River Distributions

37

39

41

43

45

47

49

51

53

55

57

59

61

63

65

4,4'-DDE

16000000

17000000

1800000019000000

20000000

21000000

2200000023000000

24000000

25000000

2600000027000000

28000000

29000000

3000000031000000

32000000

33000000

Iron

5000

6000

7000

8000

9000

Fluoranthene

1000

1100

1200

1300

1400

1500

1600

1700

1800

1900

2000

2100

2200

2300

2400

2500

Mercury

150000

160000

170000

180000

190000

200000

210000

220000

230000

240000

250000

Lead

0.3

0.4

0.5

0.6

2,3,7,8-TCDD

50000000

60000000

70000000

80000000

90000000

TOC

1000

1100

1200

1300

1400

1500

1600

TPCB

0.4

0.5

0.6

0.7

0.8

Total Tetra Dioxin

gamma-Chlordane


Figure A-2bMonte Carlo Simulation of 10,000 Iterations of Chemical Profile for Newark Bay Assuming Bounded Normal Distributions

700

900

1100

1300

1500

1700

1900

2100

2300

2500

2700

2900

3100

Benzo(a)pyrene

3

4

5

6

7

8

9

10

11

12

13

trans-Chlordane

90000

100000

110000

120000

130000

140000

150000

160000

Chromium

90000

100000

110000

120000

130000

140000

150000

160000

170000

180000

190000

Copper

Newark Bay Distributions

10

20

30

40

50

4,4'-DDE

24000000

25000000

26000000

27000000

28000000

29000000

30000000

31000000

32000000

33000000

34000000

35000000

36000000

37000000

38000000

39000000

Iron

2000

3000

4000

Fluoranthene

2000

3000

4000

Mercury

80000

90000

100000

110000

120000

130000

140000

150000

160000

170000

180000

190000

200000

Lead

0.04

0.05

0.06

0.07

0.08

0.09

0.1

0.11

0.12

0.13

2,3,7,8-TCDD

20000000

30000000

40000000

50000000

TOC

400

500

600

700

800

TPCB

0.10.110.120.130.140.150.160.170.180.190.2

0.210.220.230.240.250.260.270.280.29

Total Tetra Dioxin

gamma-Chlordane


Figure A-2cMonte Carlo Simulation of 10,000 Iterations of Chemical Profile for Upper Passaic River Assuming Bounded Normal Distributions

4 0 0 0

5 0 0 0

6 0 0 0

7 0 0 0

8 0 0 0

9 0 0 0

B e n z o (a )p y re n e

2 0

3 0

4 0

tra n s-C h l o rd a n e

2 0 0 0 0

3 0 0 0 0

4 0 0 0 0

5 0 0 0 0

C h ro m iu m

5 0 0 0 0

6 0 0 0 0

7 0 0 0 0

8 0 0 0 0

C o p p e r

U p p e r P a ssa ic D istr ib u tio n s

8

10

12

14

16

18

20

22

24

26

4,4 '-D D E

130 000 00140 000 00

150 000 00160 000 00170 000 00

180 000 00190 000 00

200 000 00210 000 00220 000 00

230 000 00

I ro n

600 0

700 0

800 0

900 0

100 00

110 00

120 00

130 00

140 00

150 00

160 00

F lu o ra n th e n e

5 006 007 00

9 001 0001 1001 200

1 4001 5001 6001 700

M e rcu ry

90000

100000

110000

120000

130000

140000

150000

160000

Lead

0.001

0.0012

0.0014

0.0016

0.0018

0.002

0.0022

0.0024

0.0026

0.0028

2,3,7,8-TCDD

30000000

40000000

50000000

60000000

70000000

TOC

300

400

500

600

700

TPCB

0.03

0.04

0.05

0.06

0.07

Tota l Tetra Dioxin

gamma-Chlordane


Figure A-2dMonte Carlo Simulation of 10,000 Iterations of Chemical Profile for Saddle River Assuming Bounded Normal Distributions

260028003000320034003600380040004200440046004800

B e n z o (a )p yre n e

40

50

60

70

tra n s-C h lo rd a n e

150001700019000

2100023000

250002700029000

3100033000

3500037000

C h ro m iu m

30000

40000

50000

60000

70000

C o p p e r

S a d d le R ive r D istrib u tio n s

1 3

1 5

1 7

1 9

2 1

2 3

2 5

2 7

4 ,4 '-D D E

9 0 0 0 0 0 0

1 0 0 0 0 0 0 0

1 1 0 0 0 0 0 0

1 2 0 0 0 0 0 0

1 3 0 0 0 0 0 0

1 4 0 0 0 0 0 0

1 5 0 0 0 0 0 0

1 6 0 0 0 0 0 0

1 7 0 0 0 0 0 0

1 8 0 0 0 0 0 0

I ro n

6 0 0 0

7 0 0 0

8 0 0 0

9 0 0 0

1 0 0 0 0

1 1 0 0 0


7 0

8 0

9 0

1 0 01 1 0

1 2 0

1 3 0

1 4 0

1 5 01 6 0

1 7 0

M e rc u ry

40000

50000

60000

70000

80000

90000

Le a d

0.002

0.003

0.004

0.005

0.006

2,3,7,8-TCDD

30000000

40000000

50000000

60000000

70000000

80000000

TO C

100200

300400500

600700

800900

1000

1100

TP CB

0.019

0.021

0.023

0.025

0.027

0.029

0.031

0.033

Tota l Te tra Diox in

gamma-Chlordane


Figure A-2eMonte Carlo Simulation of 10,000 Iterations of Chemical Profile for Second River Assuming Bounded Normal Distributions

1000

2000

3000

4000

5000

6000

7000

Be nzo(a )pyre ne

20

30

40

50

tra ns-Chlorda ne

180002000022000240002600028000300003200034000360003800040000

Chrom ium

0

10000

20000

30000

40000

50000

60000

Coppe r

Se cond Rive r Distributions

10

20

30

40

50

4,4'-DDE

13000000

14000000

15000000

16000000

Iron

2000

4000

6000

8000

10000

12000

14000

16000

18000

Fluora nthe ne

100

200

300

400

M e rcury

100000

120000

140000

160000

180000

200000

220000

240000

260000

Le ad

0.0001

0.0003

0.0005

0.0007

0.0009

0.0011

0.0013

0.0015

0.0017

2,3,7,8-TCDD

20000000

30000000

40000000

50000000

60000000

70000000

80000000

90000000

100000000

TOC

405060708090

100110120130140150160

TPCB

0.003

0.005

0.007

0.009

0.011

0.013

0.015

0.017

0.019

0.021

0.023


gamma-Chlordane


Figure A-2fMonte Carlo Simulation of 10,000 Iterations of Chemical Profile for Third River Assuming Bounded Normal Distributions

380039004000410042004300440045004600470048004900

Be nzo(a )pyre ne

40

50

60

70

80

90

100

110

120

130

140

tra ns-Chlorda ne

27000

29000

31000

33000

35000

37000

39000

41000

43000

Chrom ium

50000

60000

70000

80000

90000

100000

Coppe r

Third Rive r Distributions

3 53 73 94 14 34 54 74 95 15 35 55 75 9

4 ,4 '-D D E

1 00 0 00 00

1 20 0 00 00

1 40 0 00 00

1 60 0 00 00

1 80 0 00 00

2 00 0 00 00

2 20 0 00 00

2 40 0 00 00

I ro n

9 00 0

1 00 0 0

1 10 0 0


3 00

4 00

5 00

6 00

M e rcu ry

130000

140000

150000

160000

170000

180000

Le a d

0.0017

0.0018

0.0019

0.002

0.0021

0.0022

0.0023

0.0024

2,3,7,8-TCDD

40000000

50000000

60000000

70000000

80000000

TOC

100

200

300

400

500

600

700

800

900

1000

TPCB

0.013

0.015

0.017

0.019

0.021

0.023

0.025

0.027

0.029

0.031


gamma-Chlordane


Figure A-2gMonte Carlo Simulation of 10,000 Iterations of Chemical Profile for CSO Assuming Bounded Normal Distributions

1 0 0 0

2 0 0 0

3 0 0 0

4 0 0 0


0

1 0

2 0

3 0

4 0

5 0

6 0

7 0


1 0 0 0 03 0 0 0 05 0 0 0 0

7 0 0 0 09 0 0 0 0

1 1 0 0 0 01 3 0 0 0 01 5 0 0 0 0

1 7 0 0 0 01 9 0 0 0 0

2 1 0 0 0 02 3 0 0 0 0

C h ro m iu m

1 0 0 0 0 0

2 0 0 0 0 0

3 0 0 0 0 0

4 0 0 0 0 0

5 0 0 0 0 0

6 0 0 0 0 0

7 0 0 0 0 0

8 0 0 0 0 0

C o p p e r

C S O D istr ib u tio n s

10

20

30

40

50

4,4 '-D D E

10000000

20000000

30000000

40000000

50000000

I ro n

100020003000400050006000700080009000

10000110001200013000


1000

2000

3000

M e rcu ry

100000200000300000400000500000600000700000800000900000

10000001100000120000013000001400000

Lead

00.0020.0040.0060.008

0.010.0120.0140.0160.018

0.020.0220.024

2,3,7,8-TCDD

200000000

300000000

400000000

500000000

TOC

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

TPCB

0

0.1

0.2

0.3

0.4

0.5


gamma-Chlordane


Figure A-2hBootstrap Simulation of 10,000 Iterations of Chemical Profile for Resuspension Source Based on 1995 TSI 0-6 inch Data (TSI Location

246 Excluded)

2 0 0 0

3 0 0 0

4 0 0 0


1 9

2 1

2 3

2 5

2 7

2 9

3 1

3 3

3 5


1 2 0 0 0 0

1 3 0 0 0 0

1 4 0 0 0 0

1 5 0 0 0 0

1 6 0 0 0 0

1 7 0 0 0 0

1 8 0 0 0 0

1 9 0 0 0 0

C h ro m iu m

1 7 0 0 0 0

1 9 0 0 0 0

2 1 0 0 0 0

2 3 0 0 0 0

2 5 0 0 0 0

2 7 0 0 0 0

2 9 0 0 0 0

3 1 0 0 0 0

3 3 0 0 0 0

3 5 0 0 0 0

3 7 0 0 0 0

C o p p e r

R e su sp e n sio n D istr ib u tio n s

5 0

6 0

7 0

8 0

9 0

4 ,4 '-D D E

2 2 0 0 0 0 0 0

2 3 0 0 0 0 0 0

2 4 0 0 0 0 0 0

2 5 0 0 0 0 0 0

2 6 0 0 0 0 0 0

2 7 0 0 0 0 0 0

I ro n

4 0 0 0

5 0 0 0

6 0 0 0

7 0 0 0

8 0 0 0

9 0 0 0

1 0 0 0 0


2 6 0 0

2 8 0 0

3 0 0 0

3 2 0 0

3 4 0 0

3 6 0 0

3 8 0 0

4 0 0 0

4 2 0 0

M e rcu ry

270000280000290000300000310000320000330000340000350000360000370000380000390000

Lead

0.3

0.5

0.7

0.9

1.1

1.3

1.5

1.7

1.9

2,3,7,8-TCDD

80000000

90000000

100000000

110000000

120000000

TOC

1400

1600

1800

2000

2200

2400

2600

2800

3000

3200

TPCB

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2


gamma-Chlordane


Figure A-3Monte Carlo Simulation of 10,000 Iterations of Decay of Excess Sediment Contamination () Assuming Bounded Normal Distribution

-0.03

-0.02

-0.01

0

0.01

Chlordane

-0.024

-0.022

-0.02

-0.018

-0.016

-0.014

-0.012

-0.01

-0.008

-0.006

-0.004

Copper

-0.05

-0.04

-0.03

DDE

Lamda Distributions

-0.04

-0.03

-0.02

-0.01

Diox

-0.032-0.03

-0.028-0.026-0.024-0.022-0.02

-0.018-0.016-0.014-0.012-0.01

Lead

-0.03

-0.02

-0.01

Mercury

-0.04

-0.03

-0.02

PCB

gamma-Chlordane 4,4’-DDE

2,3,7,8-TCDD TPCB


Figure A-4Bootstrap Simulation of 10,000 Iterations of Net Sediment Rate Based on the Differences Between the 1989 and 2007 Bathymetric Surfaces

0.25

0.26

0.27

0.28

0.29

0.3

0.31

0.32

0.33

Net Sediment Rate (in/yr)


Figure A-5Simulation of 10,000 Depth of Mixed Layer Based on Random Variation Between 10 and 20 cm

10

11

12

13

14

15

16

17

18

19

20

Depth(cm)

ATTACHMENT B

ESTIMATING THE COMMON HALF TIME FOR LEGACY SEDIMENTS

IN LOWER PASSAIC RIVER

Attachment B: Estimating the Common Half Time for Legacy Sediments in Lower Passaic River.

1.0 Summary

A first-order regression model was applied to the excess chemical concentrations1 and estimated time of deposition in the Lower Passaic River in order to determine a common half-time for legacy contaminated sediments. The data used in the model came from high resolution cores collected in the Lower Passaic River, and concentrations observed for the external sources. The chemicals included in the model were: gamma-Chlordane, 2,3,7,8-TCDD, Total PCB, 4,4’-DDE, mercury, lead, and copper. The results of the analysis indicate a common decay process2

for these sediments at an average half time of approximately 35 years. The 95 percent confidence interval for this common half time is from 27 to 48 years. Although only seven chemicals were included in the model, this result also applies to other particle reactive contaminants in the Lower Passaic River that have a significant resuspension source term.

2.0 Objectives • Determine whether the chemical specific decay rates or half times on the excess

concentrations are similar (i.e., no significant difference amongst them). • Estimate the common decay rate for the excess concentrations in legacy sediment in the

Lower Passaic River, along with the associated confidence interval.

3.0 Methods • The chemicals included in the analysis were: gamma-Chlordane, 2,3,7,8-TCDD, Total

PCB, 4,4’-DDE, mercury, lead, and copper. • High-resolution core and external source data were used in the analysis. For the high

resolution cores, data were limited to segments with approximate years from 1980 to 2007.

• A multiple-regression analysis was conducted to determine the similarities and differences amongst the half times of the various chemicals. This model combined the excess concentrations and time of deposition for all the chemicals. In addition, it included indicator variables for the chemical type and allowed for interaction effects between deposition time and chemical type. The first-order regression model used was:

1 Excess chemical concentrations were defined as the Lower Passaic River sediment concentrations less the concentrations from the external sources. 2 The term decay is used here to quantify the net processes that result in the decline of chemical concentrations over time as observed in the high resolution cores. Attachment B: Estimating the Common Half Time for 2014 Legacy Sediments in Lower Passaic River B-1 Lower Eight Miles of the Lower Passaic River

loge ExCi = β0 + β1Ti + β2Chli, + β3PCBi + β4DDEi + β5Hgi + β6Cui + β7Pbi + β8TiChli + β9TiPCBi + β10TiDDEi + β11TiHgi + β12TiCui + β13TiPbi + εi

Where:

loge ExCi = natural logarithm of the excess chemical concentrations (i.e., high resolution core concentrations less external levels from head of tide, tributaries and CSO/SWOs).

β0…β13 = regression coefficients.

Ti = estimated deposition time from high resolution core dating.

Chli = indicator variable = 1 if chemical is gamma-Chlordane, 0 otherwise.

PCBi = indicator variable = 1 if chemical is Total PCB, 0 otherwise.

DDEi = indicator variable = 1 if chemical is 4,4’-DDE, 0 otherwise.

Hgi = indicator variable = 1 if chemical is mercury, 0 otherwise.

Cui = indicator variable = 1 if chemical is cupper, 0 otherwise.

Pbi = indicator variable = 1 if chemical is lead, 0 otherwise.

TiChli, TiPCBi, TiDDEi, TiHgi, TiCui, TiPbi = interactions effects between time of deposition and chemical type.

Although there are seven chemicals, only six indicators were included (the indicator variable for 2,3,7,8-TCDD was not included). In the statistical theory of qualitative predictor variables, a qualitative variable of “c” classes is always represented by “c-1” indicator variables to avoid computational difficulties. In this application, the regression for 2,3,7,8-TCDD can be represented by all other indicator values being equal to zero. Note that the exclusion of the 2,3,7,8-TCDD does not affect model results. If the indicator variable of any the other chemicals modeled was excluded, the same regression results will be obtained.

• If the regression coefficients of the interaction terms are not statistically significant, then it can be concluded that the regression lines between natural logarithm of excess concentrations versus time for the individual chemicals are parallel, and that a common decay process occurs.

Attachment B: Estimating the Common Half Time for 2014 Legacy Sediments in Lower Passaic River B-2 Lower Eight Miles of the Lower Passaic River

4.0 Results Table B-1 presents the regression output for the first order model described above. A statistically significant model was obtained (p < 0.001 from Analysis of Variance results). The most important finding from this regression analysis is that the interaction terms are not significant (p > 0.05). Therefore, the individual chemical regressions are parallel and there is a common decay process for the legacy contaminated sediments in the Lower Passaic River. This legacy sediment represents the resuspension source that is the dominant contribution for most chemicals. Note that the residuals of this regression satisfy the regression assumptions of normality and homogeneity of variance.

Given that a common decay process exist for the Lower Passaic River excess legacy chemical concentrations, a second regression run was conducted to estimate the common decay rate and corresponding half time. For this regression run, the interaction terms which are not statistically significant were dropped from the regression equation. Table B-2 and Figure B-1 present the results for this reduced regression output. This reduced model and all the regression coefficients are statistically significant (p < 0.0001), and the chemical specific regressions lines are approximately parallel. The residuals of this reduced regression satisfy the regression assumptions of normality and homogeneity of variance. The regression coefficient for the time of deposition (β1) under the reduced regression model, which represents the common decay rate is -0.02 (Table B-2). This common decay rate corresponds to a half time of approximately 35 years. Using the standard error and t-values from Table B-2 for β1, the 95 percent confidence interval for β1 is -0.026 to -0.014. The corresponding common half time confidence interval is 27 to 48 years.


Table B-1: Regression results with interaction terms


Table B-2: Regression results without interaction terms



Figure B-1Illustration of natural logarithm of observed excess chemical concentration, time of deposition and fitted Regression Function

ATTACHMENT C

DERIVATION OF THE TRAJECTORIES

Attachment C: Derivation of the Trajectories The EMB model and the trajectory calculations are based on the assumption that the contaminant load in the Lower Passaic River is carried to the river on the incoming suspended sediment and that the masses contributed by each source are additive. Based on this assumption, we can develop a mass balance equation for each contaminant like Equation C-1.

CSOCSORRRRSRSRDDDDNBNBRSPRSPTT SCSCSCSCSCSCSCSC ++++++= 2233 Equation C-1

where CT = contaminant concentration on recently deposited sediment (Be-7 bearing) CRSP = contaminant concentration on resuspended sediment within the Lower

Passaic River CNB = contaminant concentration on sediment entering from Newark Bay CDD = contaminant concentration on sediment entering over Dundee Dam (from

the Upper Passaic River) CSR = contaminant concentration on sediment entering from Saddle River C3R = contaminant concentration on sediment entering from Third River C2R = contaminant concentration on sediment entering from Second River and

SWOs CCSO = contaminant concentration on sediment entering from CSOs ST = total solids load in the Lower Passaic River SRSP = total solids load from resuspension within the Lower Passaic River SNB = total solids load delivered from Newark Bay SDD = total solids load delivered over Dundee Dam (from the Upper Passaic

River) SSR = total solids load delivered from Saddle River S3R = total solids load delivered from Third River S2R = total solids load delivered from Second River and SWOs SCSO = total solids load delivered from CSOs If we define fi as the fraction of the total solids load in the Lower Passaic River originating with source, i, we can rewrite Equation C-1 as follows:

CSOCSORRRRSRSRDDDDNBNBRSPRSPLPR fCfCfCfCfCfCfCC ++++++= 2233 Equation C-2

where CLPR = concentration of recently deposited sediment in the Lower Passaic River fRSP = fraction of the Lower Passaic River solids load originating as

resuspension within the river fNB = fraction of the Lower Passaic River solids load originating in Newark

Bay fDD = fraction of the Lower Passaic River solids load originating in the Upper

Passaic River fSR = fraction of the Lower Passaic River solids load originating in the Saddle

River

Attachment C: Derivation of Trajectory C-1 2014 Lower Eight Miles of the Lower Passaic River

f3R = fraction of the Lower Passaic River solids load originating in the Third River

f2R = fraction of the Lower Passaic River solids load originating in the Second River or the SWOs

fCSO = fraction of the Lower Passaic River solids load originating in the CSOs The purpose of the EMB model was to find the best set of fractions for each of the solids sources to balance Equation C-2 for all of the contaminants of concern. The mechanics of the EMB model and the model results are described in more detail in Appendix F of the FFS. Table C-1 below shows the resulting fractions for each source based on the best solution to Equation C-2. Table C-1: EMB Results – Fractional Contributions of Solids from each Source Solids source Percent contribution Resuspension 47.8% Newark Bay 13.5% Upper Passaic River 32.3% Saddle River 3.6% Third River 0.5% Second River and SWOs 1.3% CSOs 0.3% In order to project the contaminant concentrations in the river (CLPR), we need to examine the past behavior of each component of the contaminant loads. In order to avoid including any effects of turning off sources, we will only examine the contaminant histories back to 1980. There is no data available to quantify the past behavior of any of the tributaries or the CSOs and SWOs. Since these sources represent less than 5 percent of the total solids load, they cannot represent a significant portion of any contaminant load. Thus, their future trajectories are not important to this process and we will assume they are constant. An understanding of the past behavior of sediment concentrations in the Upper Passaic River is based on a single high-resolution core from Dundee Lake, which was extracted by scientists at Renssalaer Polytechnic Institute (RPI) and analyzed by the USEPA. This core is described in more detail in Data Evaluation Report No. 3 - Contaminant History as Recorded in the Sediments in Appendix A. For all of the contaminants to be projected except HMW PAHs, the Dundee Lake sediments generally decreased in concentration from 1980 to about 1985 or 1990. After 1990 all contaminants (except HMW PAHs) are generally constant. We can use this information to project the Upper Passaic River component into the future as a constant source. For HMW PAHs, the concentrations seem to decrease from 1980 to the mid 1990s and then they begin to increase. That increase may continue in the future, but for the purposes of this analysis, we will assume a constant concentration from 2005 into the future. This means that the trajectory analysis may be under-predicting the HMW PAHs concentrations in the future.


The Upper Passaic River component of the Lower Passaic River sediments is not assumed to have been constant in the past, however. The data points from the RPI Core were linearly interpolated to provide a concentration for each year from 1980 to 2005 and the actual measured values were used in the analysis. Newark Bay and the resuspension term are both assumed to be declining, but without any specific information on either source, they cannot easily be separated. Therefore, both were assumed to decay exponentially with the same half time (time required for the concentration to drop by 50 percent). The half time was calculated by separating the constant sources and the Dundee Dam source from the total concentrations reported on the dated high resolution cores from the Lower Passaic River according to the following equation:

CSOCSORRRRSRSRDDDDHRCEX fCfCfCfCfCCC −−−−−= 2233 Equation C-3

where CEX = “excess” concentration (originating from Newark Bay and resuspension) CHRC = concentration reported on the high resolution cores The concentrations for the tributaries and the CSOs were based on averages from the measurements taken during the 2007/2008 sampling events. The concentration for the Upper Passaic River was based on a linear interpolation of data from the dated RPI core from Dundee Lake. The resulting data was fit to an exponential decay curve, defined by Equation C-4.

toeCtC λ−=)( Equation C-4

where C(t) = concentration at time, t Co = concentration at t = 0 λ = decay coefficient t = time The half time of the decay curve can be calculated from the decay coefficient as shown in Equation C-5.

( )λ

21ln

21 =t Equation C-5

where t1/2 = half time for the exponential decay curve The next step is to calculate the concentrations of sediments from each source into the future. For the constant sources (tributaries and SWOs) and the Upper Passaic River, this is easy. These sources are assigned the average concentration from recent sediment samples from 2005 throughout the length of the trajectory. Newark Bay and resuspension contributions must be calculated using the decay rate defined above. Attachment C: Derivation of Trajectory C-3 2014 Lower Eight Miles of the Lower Passaic River

Equation C-4 shows that the limit of the concentration as time approaches infinity is zero. Although we are assuming an exponential decay for Newark Bay, we do not expect the Newark Bay sediment concentrations to reach zero. The EMB model used the northern samples from Newark Bay as the end members. Most contaminants show an increasing trend in concentration from the south to the north end of the bay, probably indicating mixing with Passaic River sediments. For this analysis, we assume that the Newark Bay sediments will never drop below the level recently measured in the southern samples in Newark Bay. This floor value is implemented as shown in Equation C-5, which forces the concentration to approach CFloor as time approaches infinity.

( ) ( )[ ] ( )Floor

tFloorNBNB CeCCtC +−= −− 20052005 λ Equation C-5

where CFloor = minimum concentration on Newark Bay sediments as defined by southern

samples There is also a floor value associated with the resuspended sediment concentrations. Because the resuspending sediment is actually the reworking of old sediment, its concentration cannot drop beneath a floor which is defined as the sum of all other sources. Equation C-6 shows the calculation of the floor for any contaminant. The southern Newark Bay sediments were again used to define the Newark Bay contribution.

CSORRSRDDNB

CSOCSORRRRSRSRDDDDNBNBFloor ffffff

fCfCfCfCfCfCC+++++

+++++=

23

2233 Equation C-6

where CFloor = minimum concentration on resuspended sediments Similar to Equation C-5, the resuspension concentration was calculated using Equation C-7. The 2005 concentration, CRSP(2005) was assigned based on the average concentration measured in the 1995 TSI surface samples.

( ) ( )[ ] ( )Floor

tFloorRSPRSP CeCCtC +−= −− 20052005 λ Equation C-7

where CFloor = minimum concentration on resuspended sediments as defined in Equation C-

6 When all of the components had been projected into the future, Equation C-2 was used to sum them up and predict the future concentration and construct the No Action (Alternative 1) trajectory. Constructing trajectories for the remediation options (Alternatives 2, 3 and 4) involved more manipulations of the data. We can simplify Equation C-2 to Equation C-8 by defining:


∑ +++++= CSOCSORRRRSRSRDDDDNBNBii fCfCfCfCfCfCfC 2233

Equation C-8

Because the volume of solids contributed by resuspension and the characteristics of those solids change with remediation, we define the remediated concentration of surface sediments in the Lower Passaic River using Equation C-9.

∑+= iiRSPRSPREM fCfCC ''' Equation C-9

where CREM = Lower Passaic River surface concentrations after remediation C’RSP = contaminant concentration on resuspended sediments after remediation f’RSP = fraction of solids originating as resuspension after remediation f’i = fraction of solids originating with source, i, after remediation Using the definition of f, we can convert Equation C-9 to Equation C-10.

T

ii

T

RSPRSPREM S

SCS

SCC''

'' ∑+= Equation C-10

where S’RSP = solids load contributed from resuspension after remediation Si = solids load contributed from source, i S’T = solids load in the Lower Passaic River after remediation We define the following variables: a = the unremediated fraction of the erosional silt areas of the river bed b = the remediated fraction of the erosional silt areas of the river bed a = 1-b c = the amount of resuspension occurring in the remediated areas per unit area as a

fraction of the resuspension occurring in unremediated areas per unit area Using these definitions, we can define the solids load from resuspension after remediation according to Equation C-11.

( ) ( )( )cbSaSS RSPRSPRSP +=' Equation C-11 The first term on the right-hand side of the equation represents the resuspension occurring in the unremediated areas of the river, while the second term includes the resuspension occurring in the remediated areas of the river. This equation can be simplified to Equation C-12.

∑+= iiRSPRSPLPR fCfCC


( )( )cbSS RSPRSP −−= 11' Equation C-12 The total solids load in the river after remediation can be defined by Equation C-13.

RSPRSPTT SSSS '' +−= Equation C-13 By plugging Equation C-13 into Equation C-12 and simplifying, we get Equation C-14.

( )( )cbSSS RSPTT −−= 1' Equation C-14 We can simplify Equation C-14 further by using the definition for fRSP.

( )( )[ ]T

TRSPTT S

ScbSSS −−= 1'

( )( )( )cbfSS RSPTT −−= 11' Equation C-15 The concentration of the resuspending sediment after remediation is defined by Equation C-16.

( ) ( )( )bca

cbCaCC capRSP

RSP +

+=' Equation C-16

where Ccap = concentration of sediment resuspending from remediated areas As described in Appendix C of the FFS and the Data Evaluation Reports, there is no appreciable dioxin contamination above RM12, so the fraction of solids eliminated by remediating is not the same as the fraction of dioxin mass isolated by remediating. To account for this issue, we define a new variable d, as the difference between a and the fraction of dioxin mass isolated by remediating. For all other contaminants, d will remain zero. This addition to Equation C-16 yields Equation C-17.

( ) ( )( )bca

cbCdaCC capRSP

RSP +

+−=' Equation C-17

where d = the difference between the fraction of solids eliminated by remediating and

the fraction of dioxin mass isolated by remediating Equation C-17 reduces to Equation C-18 using the definition of a.

( ) ( )( )( )cb

cbCdbCC capRSP

RSP −−+−−

=11

1' Equation C-18


The final equation for the remediated concentration of suspended sediment in the Lower Passaic River can then be developed by substituting Equations C-12, C-15 and C-18 into Equation C-10 as shown in Equation C-19.

( ) ( )( )( ) ( )( )[ ]

( )( )( ) ( )( )( )cbfSfC

cbfS

cbScb

cbCdbC

CRSPT

ii

RSPT

RSPcapRSP

REM −−+

−−

−−

−−+−−

= ∑1111

1111

1

Equation

C-19

Through additional simplification and remembering the definition of fRSP and fi, Equation C-19 can be converted to Equation C-20.

( ) ( )( )( )( )( ) ( )( )cbf

fCcbf

cbCdbCfC

RSP

ii

RSP

capRSPRSPREM −−

+−−

+−−= ∑

11111

Equation C-20

The calculation of Ccap is based on the assumption that there is a 6-inch biologically active zone that is well-mixed and available for resuspension. The average annual sedimentation rate was determined to be 0.27 inches. Based on this data, we can estimate that it takes 22 years to build up a 6-inch layer in the remediated area. Ccap can then be calculated by assuming 5.73 inches have the same concentration as Ccap the previous year and 0.27 inches have the same concentration as the newly deposited sediment (CREM) the previous year. The two pieces are assumed to be completely mixed and Ccap can be calculated as shown in Equation C-21.

( ) ( ) ( )[ ]1121

221

−−+= tREMtcaptcap CCC Equation C-21

where t = time (in years) Equation C-21 is also used to calculate the average concentration in the upper six inches of the sediment in the remediated areas. For the No Action Alternative, the average concentration in the upper six inches is estimated in much the same way, except the 1995 value is estimated as the average of the previous 22 years of Be-7 bearing sediment concentrations. The final necessary calculation is c, which is defined as the amount of resuspension occurring in the remediated areas per unit area as a fraction of the resuspension occurring in unremediated areas per unit area. During remediation, resuspension is assumed to decline linearly until remediation is complete. For Alternatives 2 and 3, immediately after remediation, c is assumed to be zero since the cap or backfill will be made up of sand particles, too large to be suspended under normal hydrodynamic conditions. However, as time passes, more and more sediment will be deposited in the remediated area. At some point, resuspension in the river is assumed to reach the level it had before remediation. For purposes of this analysis, we assume that this level will be reached once the river has deposited a full 6-inch biologically active layer. Using the average annual deposition rate


of 0.27 inches, this point will be reached 22 years after remediation. For simplicity, the value of c is assumed to move linearly from a value of zero just after remediation is complete (i.e., 2029 for Alternative 2, 2022 for Alternative 3) to a value of one in the year 2051 for Alternative 2 and 2044 for Alternative 3. For Alternative 4, the mixing described in Equation C-21 is also used to calculate the concentration in the upper 6-inch biologically active zone for the unremediated areas below RM8.3. The area averaged concentration across both remediated and unremediated areas was then calculated by assuming one-third of the area below RM8.3 was remediated.


Appendix C Mass Balance Modeling Analysis - Passaic River ...

Documents

Transcript of Appendix C Mass Balance Modeling Analysis - Passaic River ...